Cereals_Cost_benefit_Analysis.pdf

COST-BENEFIT ANALYSIS OF A CEREAL

VALUE CHAIN IN ETHIOPIA

(AGP-AMDe PROJECT)

FINAL REPORT

Cost-Benefit Analysis of Cereal Value Chain in Ethiopia, December 2012

UNITED STATES AGENCY FOR INTERNATIONAL DEVELOPMENT

LEARNING, EVALUATION, AND ANALYSIS PROJECT

(AID-OAA-C-11-00169)

COST-BENEFIT ANALYSIS OF CEREAL VALUE CHAINS IN ETHIOPIA

(AGP-AMDe PROJECT)

FINAL REPORT

Prepared for:

United States Agency for International Development/Ethiopia Attn: Mark Carrato, Office of Economic Growth and Transformation, USAID/Ethiopia

Prepared by:

Optimal Solutions Group, LLC Cambridge Resources International, Inc.

University of Maryland Research Park, M Square

60 Montgomery Street Cambridge, MA 02141

5825 University Research Court, Suite 2800 USA College Park, MD 20740

USA

www.cri-world.com www.OptimalSolutionsGroup.com

Project Analysts:

Jerrod Mason, USAID, Washington, D.C.

Richard Barichello, Senior Cost-Benefit Analyst Mikhail Miklyaev, Cost-Benefit Analyst Katarzyna Pankowska, Cost-Benefit Analyst Habtamu Fuje, Local Consultant

December 2012

Disclaimer The authors’ views expressed in this publication do not necessarily reflect the views of the United States Agency for International Development or the United States Government.

Table of Contents

MAIZE VALUE CHAIN

Yields and Challenges

Value Chain Actors and Dynamics

Interventions and Spreadsheet Logic

Conclusions

WHEAT VALUE CHAIN

Introduction

Wheat: Characteristics and Market Potential

Yields and Challenges

Value Chain Actors and Dynamics

Interventions and Spreadsheet Logic

Modeling the without‐intervention scenario

Conclusions

References for Wheat Value Chain

References for Maize Value Chain

MAIZE VALUE CHAIN

Introduction

In terms of the area devoted to production and its contribution to nutrition, maize is the most important staple crop grown in Ethiopia. In the context of the Agribusiness and Market Development (Program) (AMDe) project, it represents the second largest share of investment at 24 percent of project funds (ACDI-VOCA, 2012). Maize probably has more potential for yield increase than for other staple crops and can be grown across several of Ethiopia’s agro-ecological regions, making it an important component of Ethiopia’s strategy for agricultural development (Rashid, Getnet, & Lemma, 2010).

AMDe’s focus in the maize value chain lies in several areas, including increasing access to improved inputs, such as fertilizer and seeds, particularly high-yielding hybrid varieties; increasing the efficiency and market presence of cooperatives and cooperative unions’ increasing the demand for maize by tapping animal feed and agro-processing demand; and improving postharvest storage tools and techniques to reduce losses of grain between the field and the market. The AMDe work plan places special emphasis on the role of cooperatives and cooperative unions, particularly in the market linkages portion of the value chain.

Specific interventions envisioned in the AMDe work plan include investment in maize Agricultural Research Centers (ARCs) to increase the availability of improved maize varieties, particularly of high-yield hybrids;

improved storage facilities for cooperatives and unions to reduce postharvest losses due to crop spoilage and rodent or other infestation;

professionalization training, capacity building, and business-plan development for cooperatives and Farmers Cooperative Unions (FCUs) to increase their efficiency and role in maize marketing; and direct-market linkage facilitation between consumers (food processors and feed producers) and cooperatives/FCUs through trade conference facilitation and advance contracts.

The cost-benefit analysis (CBA) performed to date on this value chain has focused on the impacts of investments in improved seed provision, improved storage facilities, and professionalization and capacity-building efforts to increase cooperative/FCU market share and efficiency. Other components of the project could be subjected to a similar analysis given further time and resource availability. Specific interventions that may be explored include mechanizing harvesting equipment to improve yield, decreasing labor requirements, and reducing on-farm losses due to manual harvesting technologies.

Maize: Characteristics and Market Potential

Maize is the most important crop grown and consumed in Ethiopia, in terms of the number of households engaged in growing it (7.6 million in 2010–11), total production (4.9 million metric tons (MT) in 2010–11), and the proportion of total calories consumed in the country (20.6 percent of calories consumed) (CSA, 2011). Ethiopia is the third-largest maize producer in Africa. Maize is primarily grown in the Amhara, Oromiya, Tigray, and Southern Nations, Nationalities, and Peoples (SNNP) regions, which together account for more than 90 percent of national maize production.

Regional maize yields in the major producing regions vary from 2.35 metric tons per hectare (MT/ha; SNNP) to 2.6 MT/ha (Oromiya), with the national average at 2.54 MT/ha. Within regions, yield averages diverge substantially—for example, in Oromiya, the range is 1.68 MT/ha (Jijiga zone) to 2.94 MT/ha (Horoduguru zone) The national yield average is high compared to that of other East African producers, but low compared with South Africa, Pakistan, and other major maize producers in Africa and Asia. Estimates of Ethiopia’s potential average maize yield range from 4.2 to more than 5 MT/ha.1

For most of the past decade, Ethiopia has essentially produced and consumed maize domestically, with few imports or exports. During this time, the domestic price of maize has generally been between the import and export parity prices of international maize, indicating that the costs of importing or exporting would outweigh the benefits (Rashid, Getnet, & Lemma, 2010). There have been brief episodes when price spikes have caused the domestic price to exceed the import parity price, but government control of foreign exchange for grain imports has prevented any significant imports of maize from occurring. Additionally, although global exports are generally not cost-effective, some opportunities for regional exports of maize exist.

However, in years when regional maize prices have been sufficiently high to overcome the costs of transport, the government has generally not allowed maize exports. For this reason, and due to the high costs of transportation in the region, Ethiopia has had little cross-border trade in maize.

Domestically produced maize is generally grown on smallholder farms, with the majority being reserved for on-farm use. Of the proportion of maize that does reach the market, the majority (80–90 percent) is sold through private traders, while the remainder is sold either directly in local markets or through the cooperative/cooperative union system.2 Due to the small quantities marketed by each farm (often less than 1 MT), aggregation costs are a very significant portion of the price differential between the farm-gate and market price of maize. No stable year-round market for maize exists, and the price of maize fluctuates significantly based on the season.

Maize, like other staples, is grown primarily during the meher season in Ethiopia, when seasonal rains allow the crop to grow without irrigation. Although some smallholders (less than 25 percent nationally) cultivate maize during the secondary belg season, the total maize production during this season is less than 10 percent of overall maize production and is generally used only to complement family maize consumption. For these reasons, the focus of this analysis is on the primary meher season (Taffesse, Dorosh, & Asrat, 2011).

1 For example, World Bank estimates used to evaluate the AGP assume an average yield potential of 5.3 MT/ha.

2 Although hard estimates of the percent of grains marketed through cooperatives are difficult to find, most value chain studies agree that they market only a small percentage of the total quantity sold, although this proportion has increased in recent years. See, for example, (Bernard, Spielman, Taffesse, & Gabre-Madhin, 2010), which estimates that less than 1 percent of staple crops was marketed through cooperatives in 2007.

10000 20000 30000 40000 50000 60000 70000

K g/ h a

2008‐10 Average Maize Yield

Yields and Challenges

As noted above, Ethiopia’s maize yields are relatively high compared with those of other East African producers, but low compared to those of other major maize producers in Africa and Asia. Yield improvements are possible through many channels, primarily through higher-yielding seed varieties (particularly hybrid seeds), better and more-frequent application of chemical fertilizers, better farm management and crop rotation, and improved postharvest management and storage.

The biggest current need in terms of improved maize seed in Ethiopia is the scaling-up of production and dissemination of high-yielding hybrid maize varieties. Maize, more so than simple-seed crops, such as wheat, benefits greatly from hybridization, and the potential for yield increases is very large if farmers adopt high-input/high-yield farming practices along with better crop management to reduce soil fertility losses. Several high-yielding varieties exist in Ethiopia and are specialized for the different climatic zones of the country (lowland, medium altitude, high altitude); however, a major constraint is currently that ARCs are not able to produce enough basic seed of these varieties to allow seed multipliers to scale up to the level required to meet farmers’ demands. A second problem is the fact that seed multipliers do not respond to market signals when determining what types and quantities of seed to grow, resulting in significant mismatches between supply and demand. This incompatibility led, for example, to a large quantity of the maize hybrid BH-660 going unsold due to shifts in farmer demand that were not anticipated by the centralized seed-supply system.3 In spite of these issues, the thriving business done by private maize-seed producers, such as Pioneer, indicates that there is a healthy demand for improved seed, provided that the channels to produce and distribute it can be cleared.

The other major challenge to increasing the quantity of maize marketed in Ethiopia is losses due to poor postharvest management and storage conditions, as total postharvest maize losses for Ethiopia have been estimated at between 16 and 27 percent in recent years (Estimated post-harvest losses, 2003-2011, 2012).

According to farmer interviews undertaken by the research team, about 10 percent of this loss likely occurs on the farm, where maize is generally stored in semi-impermeable sacks in the home, leaving it susceptible to moisture and rodent and other infestations. The remainder of losses probably occurs during transfer and storage at various other points in the value chain. Cooperatives and unions in particular, which often hold grain for significant amounts of time in anticipation of higher prices in the off-season months, likely suffer from the same loss sources as do farmers’ stockpiles.

Value Chain Actors and Dynamics

The maize value chain includes many actors. On the input side, inputs are supplied primarily through agricultural cooperatives at government fixed prices, either from domestically sourced producers (as in the case of improved seed) or government imports (as in the case of inorganic fertilizers and herbicides/fungicides). The single major exception is the Pioneer Seed company, which supplies improved hybrid seed directly to consumers at market-determined prices and services 25 to 30 percent of the market for improved seed.4 On the output side, farmers market their produce primarily through private traders, with a typical chain involving several levels of aggregation en route to wholesalers in Addis or in regional markets.

A very small amount of maize (probably no more than 5–7 percent) is marketed through local cooperatives and their parent cooperative unions (FCUs), which in turn sell either through wholesalers or, rarely, directly to processors or other consumers.

3 As related to us in interviews with private- and public-sector seed multipliers.

4 Estimate provided in an interview with Pioneer management, confirmed in interview with experts at the Agricultural Transformation Agency.

Generally speaking, limited evidence suggests that any individual actors in the maize value chain have significant market power; any power that traders or other value-chain actors do have appears to have been waning for at least a decade. Specifically, between 1997 and 2008, estimated transportation and handling costs and traders’ margins have decreased dramatically, by 80 percent according to one source (Rashid & Negassa, 2009). More generally, it is difficult to see how private traders, who number at least in the tens of thousands nationwide, could command significant market power, particularly when farmers are able to check prices through conferring with other farmers and consulting prevailing prices in local and regional markets as well as in Addis. In farmer interviews, the research team met farmers who use each of these strategies to determine when to sell, implying that, at least for some farmers, information and market power do not appear to be major problems.

One important exception to this general observation relates to the issue of access to capital. Farmers generally purchase inputs—fertilizer and sometimes seed—on credit, as their ability to finance these inputs from household cash is extremely limited. Further, farm households may take out additional loans during the lead-up to the harvest season, as household food stores and other supplies run low. This credit is largely provided by the agricultural cooperatives where the farmers get their inputs or by local traders who agree to take a quantity of the harvest as payment in kind. The end result is that farmers are often obliged to sell a portion of their product directly after harvest, at a time when the market price is less favorable, to liquidate their debts. In farmer interviews, the research team found that common practice was for farmers to sell half their marketed product immediately at a lower price, to repay debts and satisfy immediate cash needs, and then save the other half to sell in the future, when the price had increased.

Intra-annual price variation of maize can be substantial—during 2011, the retail price of maize varied from about ETB 280 per quintal (qt) in January to more than ETB 600/qt in September.5 Over 5 years, the average variation in minimum/maximum price is around ETB 120/qt (Ethiopia monthly market watch, 2012). This seasonal variation obviously coincides with the harvest season, as newly harvested maize begins to come to the market after September and reaches its peak in January, with the price gradually increasing as the availability of maize declines leading up to the next year’s harvest. Given this degree of interseasonal price variation, the logic behind storing grain for later sale is obvious. The reasons that more farmers, cooperatives, and traders do not do so have already been alluded to—namely, poor storage conditions leading to losses and liquidity constraints due to borrowing for inputs and other expenses during the cropping season.

Interventions and Spreadsheet Logic

The CBA models built for the maize value chain examine the impacts of three different interventions—one at the farm-inputs level and two at the cooperative level. These interventions align with the goals and specific interventions identified in the AMDe work plan, but increasing the specificity of the intervention parameters, particularly by identifying the individual cost components of each intervention and its likely coverage areas, would greatly improve the estimates of these models.

The interventions modeled apply primarily to maize farms and cooperatives in the Amhara and Oromiya regions, as they are the regions targeted for maize value chain interventions in the AMDe work plan. The first intervention modeled relates to increasing the availability of improved seed varieties to farms, with the expected result that yields will increase. The second intervention takes place at the cooperative/FCU level and attempts to capture the likely impacts of a suite of interventions aimed at increasing the volume of maize marketed through cooperatives (as opposed to private traders). The final intervention modeled attempts to evaluate an intervention to provide, through training and financing assistance, improved storage facilities and techniques to reduce postharvest losses. Each intervention is described in detail below.

5 Price data provided by Ethiopia Ministry of Agriculture and Rural Development.

Intervention 1: Increased Improved Seed Use

The first intervention analyzed is an investment at the level of Ethiopia’s maize ARCs and seed multipliers to increase production, and thereby availability, of improved hybrid maize seed to farmers. From interviews with ARCs and seed multipliers, these investments include purchasing improved cold-storage equipment;

capital equipment, such as vehicles, mechanical seeders, threshers, and seed-cleaning equipment; and training to improve the quality of seed produced for sale. These are general activities that we assume will be undertaken under this heading—more detail from the project implementer on specific intervention activities and costs, number of ARCs and seed multipliers reached, and so forth, would provide a better estimate of the project’s actual impacts.

The result of these investments is projected to be an increase in the volume of improved hybrid seeds available to farmers. In the base-case analysis, it is assumed that initial investments will yield a 5 percent increase in the total volume of improved seeds used by farmers in year 1 of the intervention, with a further 2 percent increase each year in years 2–5, yielding a net increase in improved seed provision of about 13.6 percent. In interviews with ARC researchers, they estimated that the investments specified could increase their ability to provide pre-basic and basic seed to multipliers by as much as 50 percent, which makes this CBA’s assumption of 13.6 percent growth relatively conservative.

However, given the fact that seed multiplication in Ethiopia does not take place in the context of an active private market, where price signals give producers the appropriate incentives to increase or decrease their supply, it is not clear that stimulating production at the ARC level will translate perfectly into increased supply to farmers. In fact, the prices set by the Ethiopian government for hybrid maize seed are significantly less than what the private market would dictate,6 reducing the incentives for producers to produce more and better seed. Additionally, the lack of attribution between seed multipliers and their product, caused by the aggregation of improved seed from several multipliers at the regional or national level, reduces the accountability of seed multipliers to adhere to high standards for their product. This problem has led to some notable failures of so-called “improved” seeds sold in the past and may engender some skepticism by farmers regarding the true yield potential for the publicly provided seed varieties.7 For these reasons, the decision to moderate these assumptions about the pass-through from the ARC and seed-multiplier levels to the farm level seems prudent. As discussed in the section on sensitivity analysis, the expected value of the project is sensitive to this assumption—if the project succeeds in increasing improved seed production by a larger amount, the project benefits increase substantially. Ongoing efforts to monitor increases in improved seed production will help determine whether the project is achieving the results estimated here.

Increased use of improved seed is modeled here as an increase in the proportion of improved seed used by farms relative to recycled seed. In reality, farmers generally purchase improved seed at intervals of (usually) several years, from traders, cooperatives, or other farmers, and use their own recycled seed in the interim years. This method of seed recycling generally reduces the yield of the seed, particularly in the case of hybrid seeds, which can lose as much as 30 percent of their yield potential each year that they are recycled.

Additionally, poor seed cleaning and storage techniques also reduce the yield of recycled seed. This model associates the use of recycled seed with an 18 percent reduction in yield for each year that the seed is recycled.

Again, this is probably a conservative estimate, as higher yield losses would increase the benefit of recycling seed less frequently. In the with-project scenario, it is assumed that farmers will purchase new improved seed every 2 years, as opposed to the without-project scenario of purchasing new seed every 4 years.

6 This is indicated clearly by the fact that Pioneer sells high-quality hybrid varieties at two to three times the price of the Ethiopian government-owned improved varieties and is able to retain nearly one-third of the improved seed market.

7 These problems were related to us in an interview with a private seed multiplier in Bako region.

Increased improved seed use, coupled with intensified use of fertilizer and better application techniques, is expected to generate the increase in yields that it is assumed will occur due to the project. Specifically, the yield increase for maize is projected to be 43 percent higher in year 1 of the project, relative to the without-project scenario. It should be noted again that the without-project yield increases as well due to the assumed impacts of the larger AGP program. The large incremental increase in yield in this case is due to the large potential for increased yields with the proper use of hybrid seed as opposed to recycled traditional seed. After this initial yield increase, marginal improvements in techniques (seed and fertilizer application and timing, weeding, etc.) are assumed to increase yield at a rate of 1.5 percent per year under both scenarios.

Two farm models are used here to describe the activities of different scales of farming—one describes the activities of a “small farm,” defined in this case as a farm that plants .25 ha of maize. The other model describes a “medium farm,” defined as a farm that plants 1 ha of maize.8 It should be noted that these hectarages are not the total size of the farm; in fact, interviews with farmers in different regions indicated that they often plant several different crops on their total hectarage, with one crop rarely accounting for more than 30 to 40 percent of the total hectarage. These farms, in other words, are probably more like 1 ha (for the “small farm”) and 2.5 ha (for the “medium farm”), with the remaining hectarage planted in some combination of wheat, teff, beans, or vegetables, depending on the soil and climatic conditions and the farmer’s preferences. The two farm scales are used to reflect the difference that scale allows—namely, that some farmers with “medium”-size farms use a degree of mechanization. Specifically, the research team found in farmer interviews that these farmers are more likely to rent a tractor for their first plowing to reduce the number of man-hours required for ground preparation. The model has been built to reflect this difference.

Financial Analysis

Financial analysis of the counterfactual (without-project) scenario shows that both small and medium farms are operating very close to their opportunity costs—from the farmer’s perspective, the levelized profit margin of operations (the present value of profits divided by costs) is positive (17.1 percent for small farms and 20 percent for medium farms). This means that on average, after accounting for the opportunity cost of all resources (including nonmonetized resources, such as family labor and land) the farm is, on average, earning a profit on its activities. However, this should not disguise the fact that in many years, due to poor rainfall, temperature extremes, or other conditions that reduce yields, farmers may experience a negative real return on their crops. Finally, the analysis assumes a 12 percent discount rate to account for the opportunity cost of the farmer’s resources, which may overstate the value of the farmer’s next best option; because land rights are not easily transferable, farmers’ labor is often somewhat tied to location, and their access to investment opportunities is very limited. Ultimately, a safe interpretation is that both small and medium farmers are basically getting by under the counterfactual scenario, with medium farmers doing slightly better by virtue of their use of more efficient production techniques (mechanization).

Under the project scenario, the profit margin of both small and large farmers increases, to 29 percent and

31.5 percent, respectively. Counterfactual, project, and incremental net price values (NPVs) and profit margins are displayed in Table 1, below:

It is clear that, although both small and medium farms benefit from this project, medium farms do better, because they have more land (so yield increases on a per-hectare basis impact them more) and because they were more profitable to start. The end result, evaluated over 10 years, is an increase in NPV from farm operations of ETB 2,064 (US$118) for small farms and ETB 8,247 (US$471) for medium farms.

The annual debt service coverage ratios (ADSCRs) for both small and medium farms weaken slightly with the project, as the amount by which working-capital requirements increase (due to increased purchases of fertilizer and improved seed) is more than proportional to the increase in revenues. This does not mean that the increase in revenues does not offset the higher operating costs, only that the operating costs are increasing by a larger percentage. However, among farmers who may be averse to the risk of loan default, a lower ADSCR does mean less liquidity with which to repay their loans and should be monitored to ensure that farmers are able to meet their financing needs.

1 TABLE:

FINANCIAL NPVS AND PROFIT MARGINS, FARMER PERSPECTIVE

Financial NPV, ETB

Small Farm

Medium Farm

Financial NPV, USD

Small Farm

Medium Farm

Levelized Profit Margin

Small Farm

Medium Farm

Without

Intervention With Intervention Incremental

1894 3958 2064

108 226 118

506 977 471

8854 17100 8247

Without

Intervention With Intervention Incremental

20.0% 31.5% ‐

Without

Intervention With Intervention Incremental

17.1% 29.1% ‐

2 TABLE:

FINANCIAL RATIOS, BANK PERSPECTIVE

Year 1 2 3 4 5 6 7 8 9 10

Without Intervention

Small Farm 2.64 3.28 3.33 3.39 3.45 3.51 3.56 3.62 3.68 3.74

Medium Farm 1.98 2.63 2.69 2.74 2.80 2.85 2.91 2.96 3.02 3.08

With Intervention

Small Farm 2.17 2.78 2.83 2.89 2.95 3.00 3.06 3.12 3.18 3.23

Medium Farm 1.75 2.38 2.43 2.49 2.54 2.60 2.65 2.71 2.76 2.82

Economic Analysis

The economic analysis shows that, contrary to the case with wheat, the economic returns for the maize intervention, although still positive, are lower than the financial returns (shown in Table 3). This is caused by the fact that instead of being subsidized by the government, imports of maize are taxed, which causes the market price to be higher than the true opportunity cost of providing the good. This being the case, the government is essentially losing tax revenues in this scenario, if domestic maize production is a substitute for what would otherwise be maize imports. This loss to the government is somewhat, although not fully, offset by the reduction in the foreign exchange premium that the government would otherwise lose as a result of maize imports. Additionally, the farmers receive improved seed at an artificially low price, which essentially causes a transfer from seed producers to consumers (farms). So in this case, contrary to the case of wheat, the value of the project to the farmer is actually greater than the value of the project to society, although the incremental value of both is positive. The incremental economic NPV is ETB 1,867 (US$107) for small farms and ETB 7,460 (US$426) for medium farms. The difference in the financial benefits is accounted for by the tax revenues lost by the government on maize imports as well as the losses to seed multipliers, who could charge higher prices for their product in the absence of price controls.

In scaling up the impact of the project, the total increase in improved seed is assumed to be used by small and medium farms equally, relative to their proportion of maize farms overall. So, because about 24 percent of maize hectares in Ethiopia are on “small” farms (less than 1 ha), it is assumed that about 24 percent of the

3 TABLE:

ECONOMIC NPVS

Economic NPV, ETB

Economy's Perspective

Small Farm

Medium Farm

Economic NPV, USD

Economy's Perspective

Small Farm

Medium Farm

Economic NPV including

USAID funding, USD

Economic IRR including

USAID funding

Without

Intervention With Intervention Incremental

Without

Intervention With Intervention Incremental

117 223 107

2039 3907 1867

8971 16431 7460

513 939 426

12,807,719$

175% improved seed is used by small farms, with the remaining 76 percent going to medium farms.9 Because each small farm uses a smaller quantity of seed than the medium farms, the total number of small and medium farms reached by the intervention is approximately equal.

Finally, when USAID’s investment is considered, the project’s NPV is about US$12.8 million, implying that the benefits of the project exceed the costs by this amount (see Table 3). The project’s internal rate of return (IRR) is 175 percent. Although this number is very high, the productivity increases made possible by increasing the use of improved seed suggest that it is not unrealistic.

Including the net cash flows from operations, the imputed cost of land and family wages, and the value of crops consumed on the farm, real family income under the counterfactual scenario increases, from ETB 3,265 in year 1 to ETB 4,480 in year 10 for small farmers, and from ETB 10,646 to ETB 15,570 for medium farmers over the same period. With the intervention, family incomes increase at roughly the same rate, but the initial income in year 1 is higher by ETB 550 in the case of the small farm and by ETB 2,200 in the case of the medium farm. These are real increases in household income attributable to the project. It should be restated that the income levels described here are not total household income—a typical smallholder farmer plants no more than 30 to 40 percent of his hectarage in maize or any other crop, and many households rely on one or more members working off-farm either year-round or during the dry season to augment their income. The income in this model is what the family derives as a result of maize production, not as a result of production of other crops or off-farm activities. In fact, many farmers may be affected by the project in more than one value chain—for example, farmers could get increased access to both improved wheat and maize seeds, so their income would increase from the impacts of both parts of the project. However, it is difficult to identify how much of this potential overlap exists.

9 This could potentially overstate the amount of benefit to poorer farmers, who hold less hectarage, because they may be less likely to adopt the high-input/high-output intervention.

Distributional Analysis

As mentioned in the section on economic analysis, the benefits of this project accrue primarily to farmers, due to the fact that domestic maize, unlike wheat, does not compete with subsidized imports. The losses that make up the difference between the financial and economic NPVs can be interpreted as being borne by the government in the form of reduced import-tariff revenues and by seed multipliers due to the price caps set on improved seeds.

Intervention 2: Cooperative/FCU Professionalization and Training

The second intervention considered is a suite of activities designed to improve the operating processes of farmer cooperatives and cooperative unions, linking them more directly to end consumers, such as food processors and feed plants. The eventual goal of these efforts is to increase the market share of cooperatives to increase the benefits to farmers. These benefits come primarily in the form of dividends to shareholder farmers.

It should be noted that, as in the wheat model, cooperatives and FCUs are essentially aggregated together.

The section on wheat explains in detail why this modeling choice has been made. Again, more specific information on the exact activities undertaken within this intervention area—e.g., numbers and types of trainings provided to each coop or FCU, cost per training, expected outcomes for these trainings, etc.— would increase the validity and accuracy of the model. In the absence of specific figures on costs and activities, this report uses the aggregate number identified as funding for market development and adds to it 50 percent of the budget identified for training initiatives.

Under the counterfactual scenario, it is assumed that cooperatives are currently handling 7 percent of the grains marketed. In the intervention scenario, this number increases to 15 percent, which, being more than double the current quantity, is likely as much of an increase as even the most capable cooperatives and FCUs are able to handle. In fact, given that the counterfactual scenario from the farm-level analysis is being used to determine how much member farms will be marketing in total, this represents significantly more than a doubling of the quantity marketed by FCUs now.

The operating costs of the cooperatives/FCUs are assumed to be the same between the counterfactual and intervention scenarios. In reality, it might be hoped that cooperatives that receive the intervention training and assistance may be able to increase their marketing efficiency, which could reduce their operating costs;

however, in the absence of data regarding how much these costs might come down, no change is assumed.

6 TABLE:

DISTRIBUTIONAL ANALYSIS

1.105353

7921817

6911735

14833552

‐1413805

‐612028

Stakeholder Impacts, real USD

Farmers

Small farms

Medium farms

Total

Proportion of project value accruing to farmer

Government

USAID

Intervention 3: Improved Coop Postharvest Handling and Storage

A third set of activities identified in the AMDe work plan relates to improving postharvest storage and handling to reduce losses and increase the quantity of grain marketed. This is particularly important in the case of maize, as postharvest losses for Ethiopia may be as high as 26 percent in some years (Estimated post-harvest losses, 2003-2011, 2012). The intervention modeled here is the construction of a storage warehouse to increase an FCU’s ability to store larger quantities of grain for longer periods of time. The data used to model this investment are taken from a similar-size investment in grain storage in India, as specific cost estimates for Ethiopia were not readily available (James, 2002). As the AMDe project goes forward with investments in this area, more specific data on the costs, capacity, and projected benefits of improved storage would give the model greater specificity to the Ethiopian case. In the absence of specific information on budget allocation for this type of intervention, it is assumed that 50 percent of the budget identified for use in postharvest improvements will go toward this purpose, anticipating that the resources will be used for business-plan preparation, training, and partial loan guarantees and possibly grants, all of which are identified as opportunities in the AMDe work plan.

For both interventions, it is assumed that the project will begin work with three FCUs during the first year of the project and will add two additional FCUs during each additional year. This in part reflects the targeting of four FCUs in AMDe’s year 1 work plan—it is assumed that three of these will be fully underway with projects during year 1, with the other coming on in year 2 in addition to one other FCU. Again, input from the implementer on whether these assumptions are realistic would be valuable.

The financial analysis of the counterfactual scenario shows that FCUs are making relatively small margins (about 12 percent) on their goods after accounting for operating costs and the cost of capital. The costs of operation are largely driven by the cost of grain purchased and the transportation costs incurred in moving the grain from the farm to the FCU, and then especially from the FCU to the market.

The analysis of Intervention 1 (increased coop/FCU market share) is substantially the same as that for the counterfactual scenario, because the firm’s cost structure does not change; however, increasing the quantity marketed through the FCU does increase the financial NPV of its operations by about ETB 7 million (approximately US$380,000). For Intervention 2 (improved storage facilities), the firm’s profit margin improves to about 15.5 percent, and the financial NPV for the firm increases by ETB 2.7 million (approximately US$153,000). Although in the case of wheat relatively small gains in postharvest losses made the NPV of the storage investment somewhat less attractive than the marketing intervention, in this case the large expected gains in postharvest losses due to improved storage make this intervention more attractive.

7 TABLE:

FINANCIAL NPVS AND PROFIT MARGINS, Owner's perspective

Financial NPV, ETB

Financial NPV, USD

Financial IRR

Incremental ‐

Intervention 2

5989683 12835036 8404967 6845353 2415284

Without

Intervention

Intervention 1

(Increased mkt share)

Intervention 2

(Post‐harvest storage)

Incremental ‐

Intervention 1

342268 733431 480284 391163 138016

11.85% 11.85% 15.53% ‐ ‐

A review of the firm’s financial ratios shows that they are adequate for each year. The fact that they are static is due to the construction of the working capital loan—the amount borrowed is set as a percent of annual sales. If the firm were retaining earnings as one might expect, its working-capital loan requirement might decrease as a percent of sales over time, which would cause these financial ratios to strengthen further.

However, because cooperatives and FCUs are at required (at least in theory) to return a significant portion of their profits to members as dividends, and because interviews with cooperative management and coop financers within Ethiopia showed that these organizations tend to be chronically undercapitalized, it seems likely that financial ratios will remain somewhat low. Under Intervention 2, the firm’s financial ratio strengthens as it pays off the investment loan that finances a portion of the warehouse’s construction, at which point it levels off. The ADSCRs for each scenario are favorable, indicating that the firm should be able to access capital.

Unlike the farm-level maize intervention, the economic NPV of the two FCU-level interventions are greater than the financial NPVs, indicating that some of the project’s benefits accrue to parties other than the cooperatives. In this case, the government is the other primary beneficiary in addition to the cooperatives’ members, in the form of increased tax revenues due to the increased sales of maize. The incremental benefit to society of the FCU-level intervention is ETB 12.5 million (about US$700,000) for the marketing intervention and ETB 3.5 million (US$197,000) for the postharvest-loss intervention.

In scaling up the project, the assumptions outlined above have been made: that three FCUs will be reached with the project during the first year, with two more in each subsequent year. Under these assumptions, the total number of FCUs receiving the interventions during the life of the project would be 11 under both interventions. This translates to coverage of roughly 460,000 households, assuming FCU membership of 42,000 per organization. Although identifying 11 FCUs for participation in AMDe may not be reasonable at this point, the government’s strategy of increasing the prominence of cooperatives and FCUs is likely to increase their numbers, and the 460,000 households projected to be reached is still less than 7 percent of the total number of maize-growing households in the four project regions (Amhara, Oromiya, Tigray, and SNNP), making it not unreasonable to assume that they could be serviced by a cooperative or FCU within the medium-term.

8 TABLE:

FINANCIAL RATIOS, BANK PERSPECTIVE

Year 1 2 3 4 5 6 7 8 9 10

Without Intervention ‐ 1.41 1.41 1.41 1.41 1.41 1.41 1.41 1.41 1.41

Intervention 1 ‐ 1.41 1.41 1.41 1.41 1.41 1.41 1.41 1.41 1.41

Intervention 2 ‐ 2.54 2.70 2.83 2.95 3.21 3.21 3.21 3.21 3.22

9 TABLE:

ECONOMIC NPVS

Economy's Perspective, FCU‐level

Economic NPV, ETB

Economic NPV, USD

Economic NPV including

USAID funding, USD

Economic IRR including

USAID funding

712420 197749

3,795,456$ 1,240,282$

Without

Intervention Intervention 1 Intervention 2

Incremental, Intervention 1

Incremental, Intervention 2

10908926 23376270 14369529 12467344 3460603

74% 93%

623367 1335787 821116

After scaling up the project’s impacts and applying the costs of USAID’s activities described above, Intervention 1 has a higher overall NPV (US$700,000), while Intervention 1’s economic NPV is about US$197,000. However, as is discussed in the distributional analysis of the wheat FCU interventions, when considering the distributional impacts of Intervention 1, it does not appear as attractive. Specifically, insofar as the project increases cooperative/FCU income, it does so at least partially by diverting those same resources away from private traders. The resulting loss to traders is not captured in this figure, but an attempt is made to account for it in the distributional analysis below.

As mentioned above, one problem with the economic analysis of Intervention 1 is that it does not recognize the losses that accrue to private traders, who would otherwise be benefiting from marketing the additional grain that flows through the cooperative/FCU channel. A detailed discussion of this limitation, which applies equally to both the wheat and maize value chains, is included in the distributional analysis of the wheat value chain. Using an estimate of trader gross margins of ETB 10/qt10 as the lower bound, this analysis finds the present value (PV) of losses to traders to be around US$500,000, as compared with gains to cooperatives of US$642,000. It should also be noted that this decrease in trader income is a dynamic loss, not an absolute one. The loss to traders is relative to what they would otherwise have gained in the absence of the project, as they would stand to benefit from the continued relatively small market share of cooperatives. In fact, in absolute terms, trader income as a whole should be rising as well, because increasing production due to the Agricultural Growth Program (AGP) more than compensates for their reduced market share. However, it may still be expected that traders will resist efforts to channel more grain through cooperatives, as it does represent a real economic cost to them.

10 TABLE:

DISTRIBUTIONAL ANALYSIS

Intervention 1

Cooperatives

Private Traders

Government

USAID

Intervention 2

Cooperatives

Government

USAID

Proportion of project value accruing to coop 0.55

Benefit per HH (dividend), USD 37.73

Proportion of project value accruing to coop 0.58

Stakeholder Impacts, real USD

3169363

‐642997 to ‐506119

2602953

‐1976859

Benefit per HH (dividend), USD 11.16

Stakeholder Impacts, real USD

937176

665065

‐361959

In any case, the benefits of Interventions 1 and 2 are almost evenly split between cooperatives and the government. The cooperatives gain through Intervention 1 due to increased revenue from greater volume of sales and through Intervention 2 in the form of more efficiency due to lower postharvest-loss rates; and the government gains through both interventions in the form of increased tax revenues. Finally, if coops are assumed to return 50 percent of their profits to shareholders in the form of dividends, the direct monetary benefits per household for Intervention 1 are about US$7.70 and US$9.70 per household for Intervention 2.11 As with the wheat interventions, the fact that these benefits are spread out over many households reduces their per-household impact.

Sensitivity Analysis

Sensitivity analysis was performed on several model inputs to determine how sensitive the model’s outputs are to various assumptions. At the farm level, the financial NPV of the improved-seed intervention is very sensitive to the farm-gate price for maize and to the assumption about increased yield growth—a decrease in farm-gate price below about ETB 3,250 drives the financial NPV below zero for small farms, while a farm-gate price of less than ETB 3,000 makes the NPV negative for medium farms. In both cases, however, the overall project economic NPV remains strongly positive; the same is true for the yield-increase assumption.

In addition to these, the model is sensitive to the assumption about increases in input usage over time—if input usage grows more than 1 percent per year, the financial NPV for small farms becomes negative. As with the wheat model, medium farmers are slightly less vulnerable to variation in the model’s parameters than small farmers. The value of the project from the farmers’ and economy’s perspective is not sensitive to increases in the real wage or to increases in the price of improved seed—in fact, a doubling of the price of seed from its current level would not remove farmers’ incentive to use it, indicating the degree to which improved seed is currently underpriced.

At the cooperative level, the story is broadly similar. Coop financial NPV is very sensitive to changes in both the farm-gate and sale prices for maize, which reflects the cooperatives’ very thin operating margins. They are not sensitive to changes in the assumed cost of labor for loading and unloading, although increased transportation costs would decrease the project’s benefits considerably. However, in all these cases, the economic NPV of the interventions remains strongly positive, again showing that a significant proportion of benefits from these interventions is not accruing to cooperatives but to the government.