• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Contributors to the study
 Acknowledgement
 Executive summary
 Table of Contents
 List of Tables
 List of acronyms
 Introduction
 Characteristics of program participants...
 Yield results and factors affecting...
 Financial analysis
 Economic analysis
 Sustaining the momentum
 Reference














Group Title: MSU international development working paper
Title: Green revolution technology takes root in Africa
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00086787/00001
 Material Information
Title: Green revolution technology takes root in Africa the promise and challenge of the Ministry of AgricultureSG2000 experiment with improved cereals technology in Ethiopia
Series Title: MSU international development working paper
Physical Description: xix, 66 p., map : ; 28 cm.
Language: English
Creator: Howard, Julie A ( Julie Ann )
Publisher: Dept. of Agricultural Economics, Dept. of Economics, Michigan State University
Place of Publication: East Lansing Mich
Publication Date: 1999
 Subjects
Subject: Agriculture -- Research -- Africa, Sub-Saharan   ( lcsh )
Fertilizers -- Africa, Sub-Saharan   ( lcsh )
Food crops -- Africa, Sub-Saharan   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Ethiopia
 Notes
Bibliography: Includes bibliographical references (p. 65-66).
Statement of Responsibility: Julie A. Howard ... et al..
General Note: "May 1999."
 Record Information
Bibliographic ID: UF00086787
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 44476840

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Title Page
        Page i
        Page ii
    Contributors to the study
        Page iii
        Page iv
    Acknowledgement
        Page v
        Page vi
    Executive summary
        Page vii
        Page viii
        Page ix
        Page x
        Page xi
        Page xii
        Page xiii
        Page xiv
    Table of Contents
        Page xv
        Page xvi
    List of Tables
        Page xvii
        Page xviii
    List of acronyms
        Page xix
        Page xx
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
    Characteristics of program participants compared to typical agricultural households
        Page 7
        Page 8
        Page 9
    Yield results and factors affecting crop yields and technology choice
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
    Financial analysis
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
    Economic analysis
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
    Sustaining the momentum
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
    Reference
        Page 63
        Page 64
Full Text










































MSU International
Development
Working Paper No. 76
1999


Department of Agricultural Economics
Department of Economics
MICHIGAN STATE UNIVERSITY
East Lansing, Michigan 48824
MSU Agricultural Economics Web Site: http://www.aec.msu.edu/agecon/
MSU Food Security II Web Site: http://www.aec.msu.edu/agecon/fs2/index.htm
MSU is an affirmative-action/equal-opportunity institution.


MSU International Development

Working Papers


112.019


Green Revolution Technology
Takes Root in Africa
The Promise and Challenge of the Ministry of
Agriculture/SG2000 Experiment with Improved Cereals
Technology in Ethiopia

by

Julie A. Howard, Valerie Kelly, Julie Stepanek, Eric W.
Crawford, Mulat Demeke, and Mywish Maredia


l// 2 o









MSU INTERNATIONAL DEVELOPMENT PAPERS


Carl Liedholm and Michael T. Weber

Editors

The MSU International Development Paper series is designed to further the comparative analysis
of international development activities in Africa, Latin America, Asia, and the Near East. The
papers report research findings on historical, as well as contemporary, international development
problems. The series includes papers on a wide range of topics, such as alternative rural
development strategies; nonfarm employment and small scale industry; housing and construction;
farming and marketing systems; food and nutrition policy analysis; economics of rice production
in West Africa; technological change, employment, and income distribution; computer techniques
for farm and marketing surveys; farming systems and food security research.

The papers are aimed at teachers, researchers, policy makers, donor agencies, and international
development practitioners. Selected papers will be translated into French, Spanish, or other
languages.

Individuals and institutions in Third World countries may receive single copies of papers published
since 1993 free of charge. Requests for copies and for information on available papers may be
sent to:

MSU Bulletin Office
10-B Agriculture Hall
Michigan State University
East Lansing, Michigan 48824-1039
U.S.A.

Information concerning how to purchase MSU International Development Papers is included in
the back of this publication and requests should be sent to:

MSU Bulletin Office
10-B Agriculture Hall
Michigan State University
East Lansing, Michigan 48824-1039
U.S.A.












GREEN REVOLUTION TECHNOLOGY TAKES ROOT
IN AFRICA

THE PROMISE AND CHALLENGE OF THE MINISTRY OF
AGRICULTURE/SG2000 EXPERIMENT WITH IMPROVED CEREALS
TECHNOLOGY IN ETHIOPIA







by

Julie A. Howard, Valerie Kelly, Julie Stepanek, Eric W. Crawford, Mulat Demeke, and
Mywish Maredia





May 1999


Statistical annex and copies of the questionnaire are available at
http://www.aec.msu.edu/agecon/fs2/papers/idwp76aq.pdf


This paper is published by the Department of Agricultural Economics and the Department of
Economics, Michigan State University (MSU). Funding for this research was provided by
USAID/Ethiopia and the Food Security II Cooperative Agreement (AEP-5459-A-00-2041-00)
between Michigan State University and the United States Agency for International Development,
through Africa Bureau's Office of Sustainable Development, Productive Sector Growth and
Environment Division (AFR/SD/PSGE).

Howard and Maredia are Visiting Assistant Professors, Kelly is Visiting Associate Professor,
Stepanek is a doctoral candidate, and Crawford is Professor in the Department of Agricultural
Economics, Michigan State University. Demeke is Professor in the Department of Economics,
University of Addis Ababa.












































ISSN

All rights reserved by Michigan State University. 1999

Michigan State University agrees to and does hereby grant to the United States Government a
royalty-free, non-exclusive and irrevocable license throughout the world to use, duplicate,
disclose, or dispose of this publication in any manner and for any purposes and to permit others to
do so.

Published by the Department of Agricultural Economics and the Department of Economics,
Michigan State University, East Lansing, Michigan 48824-1039, U.S.A.





ii










CONTRIBUTORS TO THE STUDY


Survey Coordinator:


Interviewers:















Data Entry Clerks:

Data Cleaning
Supervisors:


Ali Said


Survey
Supervisors:


Negusie Belete
Jemal A/Faji
Afework Dubero
Elfnesh Dubero
Zewdu Damenu
Birbo Bissa
Duresso Chimsa
Muluneh Kebede
Aynalem Mekuria
Mame Kebede
Raya Kebede
Taye Yadefa
Arkiso Masebo
Belay Kebede

Abebech Demissies


Margaret Beaver


Tsige Alemu
Tamrat Yigezu
Kebede Negusie
Aytenew Shibru
Bullo Edossa
Befekadu Tesfaye
Sebsibe Gemechu
Girma Mulugeta
Tesfaye Midikssa
Bekele Wakjira
Nigatu Bekele
Lelissa Chalchissa


Demeke Abate
Mengistu Buta
Selamyihun Kidane

Yemiserach Gebeyehu
Masresha Getachew
Belayneh A/Mariam
Biruk Seifu
Dabeta Eticha
Danil Tesfaye
Derbe Gudissa
Kumelcha Bulto
Warkaye Bekele
Mesfin Tadesse
Shirega Minuye
Techane Adugna


Belaineh Taye Beyene Taddesse


Senait Hundie


Tesfaye Zegeye Mulat Demeke


Workeneh Molla


Asrat Gebre


Drivers:








ACKNOWLEDGMENTS


The research team wishes to thank the participating farmers, the Ministry of Agriculture
Department of Extension and Cooperatives staff in Addis Ababa, East Shoa, West Shoa, and
Jimma, and the staff of SG2000 in Ethiopia for their generous assistance during the course of the
survey. We thank Habtemariam Abate, Senior Extension Expert at MOA for helping us
understand the goals, evolution and operation of the New Extension Program. Thanks go to
Techane Adugna (NFIA) and Dr. Kassahun Aberu (EAL) for providing information about
imported quantities, import and transport costs and prices of fertilizer at the import, wholesale
and retail levels. We are also grateful to Tesfaye Zegeye of the Ethiopian Agricultural Research
Organization for assistance with data cleaning and analysis, and to teff, maize, and soils
researchers from EARO who provided useful comments on the survey questionnaire and on
preliminary versions of this report. GMRP staff members were also extremely helpful: thanks to
Alemu Asfaw for his assistance with price data, to Samson Dejene for assistance with data entry
and analysis, and to Tigist Tesfaye for secretarial support. The researchers with the GMRP and
the MSU Food Security II Project who provided encouragement and willingly shared their ideas
with us on numerous survey and analysis issues are too numerous to mention individually, but
Aklu Girgre, Thomas Jayne, and Michael Weber deserve special mention.

We would also like to thank Don Beaver for assistance in the design of plot measurement
procedures and software; Rick Ward and Urs Schulthess for help in designing data collection
procedures and interpreting results for the yield determinants parts of the research; and Scott
Swinton for his assistance in designing the data analysis procedures, interpreting the results, and
helpful comments on an earlier draft of this paper.









EXECUTIVE SUMMARY


BACKGROUND AND OBJECTIVES

Ethiopia's food insecurity will increase unless it can dramatically boost agricultural productivity.
In 1993, the Sasakawa/Global 2000 Program (SG) and the Ministry of Agriculture (MOA) began
a joint program to demonstrate that substantial productivity increases could be achieved when
farmers were given appropriate extension messages and agricultural inputs were delivered on time
at reasonable prices. The program provided credit, inputs and extension assistance to participants
willing to establish half-hectare demonstration plots on their own land. In 1995, the MOA/SG
demonstration program reached more than 3,500 farmers. During the same year MOA launched
the New Extension Program (NEP) based on SG principles but managed independently. By 1997,
NEP was managing the bulk of the demonstration plots (about 650,000).

Although the MOA/SG program is widely considered to be a success, no formal analysis had been
carried out to determine its profitability. In September 1997 MOA/SG agreed to collaborate with
MSU to answer the following questions: (1) Is improved technology financially profitable for
farmers? (2) Is it economically profitable from a national perspective? (3) What factors limit crop
response to improved technologies? and (4) What challenges does the government face as it scales
up the NEP program?


METHODS

The study examined the experiences of 1997 participants and graduates of the MOA/SG program
in three zones of Oromiya Region: West Shoa (maize), Jimma (maize), and East Shoa (teff).'
Agroecological conditions in all three zones are good. Our sample included 377 farmers. Data on
yield, area and input use were collected between October and December 1997 for 226 program
plots of current participants; 107 control plots where current participants used their usual or
"traditional" practices; and 157 plots of program graduates using either improved or traditional
practices.


YIELD RESULTS AND DETERMINANTS

Maize: There is strong evidence that practices used by MOA/SG participants and graduates
resulted in much higher average yields (4.8 to 6.8 tons/ha) than those obtained on "traditional"
plots (2.8 tons/ha in Jimma), and average national and regional yields (1.9-2.1 tons/ha).



'For maize, the recommended package (0.5 ha) was 50 kg DAP, 50 kg urea, and 12.5 kg hybrid maize seed. The
teffpackage included 50 kg DAP, 50 kg urea, and 17.5 kg improved seed and herbicide.









Regrouping the survey plots to account for levels of inputs used,2 we found strong evidence that
improved seeds and fertilizer are associated with higher yields. Although farmers in West Shoa
obtained average yields as high as 3.9 tons/ha using local seed without fertilizer, farmers using
improved technologies produced an additional 2 tons/ha. Farmers using improved seed and
fertilizer at less-than-recommended rates did as well as those using higher rates, signaling the
possibility of increasing profits by slightly reducing fertilizer doses.

Despite evidence of superior average performance for the improved technologies, there was
considerable yield variability across fields within a given technology type. Although insufficient
variability and high multicollinearity across key variables made it impossible to isolate the yield
impact of seeds, DAP, and urea, we were able to identify a number of environmental and
management factors that affected yields. In Jimma, a farmer can increase yields by almost 1.5
tons/ha if he plants maize on red soils, and farmers plowing four or more times can increase yields
by 550 kg/ha. Average Jimma losses for late planting were 260 kg/ha; for incorrect row distances
they were 315 kg/ha. In West Shoa incorrect spacing reduced yields by 300 kg/ha.

Teff: Teff yields for the region average 1 ton/ha-about two-thirds of the 1.4-1.5 ton/ha average
yield obtained by sample farmers. There was no statistically significant difference between yields
on MOA/SG plots and other plot types. The same was true for teff straw. Two factors are
responsible: (1) teff farmers have been using some type of improved seed and fertilizer for many
years; and (2) MOA/SG participants were not required to use the program-recommended seed
varieties. Consequently, there were no fixed differences in input use across plot types.

When we regrouped the plots by input level we found that the plots cultivated with the MOA/SG
recommendations performed less well than those using retained seed and fertilizer at or below
recommended rates. This suggests that the variety of seed distributed with the 1997 MOA/SG
package was not as well adapted to the region as improved varieties distributed earlier.


FINANCIAL ANALYSIS

The use of improved technology was very profitable for both maize and teff farmers under a range
of output price and yield scenarios. We calculated net income per ha and per labor day using 1998
prices for January, April-May, and August to assess potential gains from storage. We also
calculated returns assuming hypothetical drops in January output prices of 25% and 50%. Finally,
we calculated returns to maize for a case where storage losses were reduced by half using storage
pesticides. Results from these six scenarios led to four key conclusions about maize and teff
profitability.

Improved technology is profitable for both maize and teff, even if output prices decline by


2 Some traditional plots used fertilizer and/or improved seed while a few graduates did not, hence regrouping the
plots by levels of inputs used made it easier to examine the impact of the technology.









25% or 50%. Net income is high for improved teff and maize under almost every price scenario.
Net income/ha ranged from 112-4406 birr/ha (6.70 birr=-USD) for farmers using improved
technology. Returns to family and mutual labor ranged from 1.6 to 48.9 birr/day for improved
technology users, far exceeding average daily wage rates (3-5 birr/day) in all cases except for teff
farmers who used program seed with fertilizer when a 50% output price drop was assumed.

In Jimma maize farmers who used the complete package of improved seed, DAP and urea
achieved double the net returns, and 40% higher returns per labor day, than farmers using the
traditional practice of local seed plus DAP. Net returns for maize farmers in West Shoa using
improved technology were 25 to 33% higher, and returns to labor were 50-60% higher, than
farmers using only local seed and no fertilizer.

Farmers who adapted technology packages-using lower fertilizer recommendations and in
the teff case a different seed type-achieved as high or higher profits than farmers who
used the full technology recommendations. Our analysis suggests the existence of a "learning
curve." After farmers' first exposure to a technology they subsequently learn more efficient ways
to apply it. Graduate farmers tended to use less labor and fertilizer than current program
participants.

Gains from storage are potentially great if storage insecticide is used and farmers are
allowed to repay loans later in the marketing year. Teff grain prices rose by 23% and straw
prices doubled between January and August. By storing and selling later, teff farmers could
increase net income by at least 40%. Maize prices also rose significantly between January and
August in West Shoa (29%) and Jimma (72%). Unlike teff, untreated maize deteriorates rapidly in
storage. In Jimma, after accounting for storage losses, net income per hectare and per labor day
rose by over 60% if farmers sold in August. In West Shoa the price rise was less dramatic, but
farmers still gained 7-8% by selling later.

None of the sample farmers reported using storage insecticide, but EARO research indicates that
use of insecticide can cut storage losses in half. Farmers using insecticide and selling in August
instead of January would increase net income by 80% in Jimma and 20% in West Shoa. Current
MOA lending policies make it difficult for farmers to store their crops for later sale as farmers are
expected to pay back loans at harvest. Unless they have other sources of income to repay loans,
most farmers would be obliged to sell when prices are lowest. Allowing farmers to pay back loans
later if they are willing to pay higher accumulated interest charges could increase returns to
improved technologies.

Improved seed and fertilizer represent 50-80% of total costs. Improved inputs are by far the
biggest cost component in the budgets. This suggests that even small reductions in the farmgate
cost of fertilizer and seed (e.g., by reducing transport and other marketing costs) could
significantly increase farm profits.









ECONOMIC ANALYSIS


The economic analysis helps answer the following questions: (1) Which is cheaper for society
overall-producing maize and teff domestically, using improved technologies, or importing
foodgrains? (2) Will it be socially profitable for Ethiopia to export maize and teff produced with
improved technology, i.e., will the economic returns from exports cover the economic costs? and
(3) Do the benefits of government-supported programs to facilitate technology adoption outweigh
their costs to society? Three key findings of the economic analysis are discussed below.

Assuming that intensified production of maize and teff fills a domestic need that otherwise
would have to be filled by commercial imports, it is highly profitable from society's
standpoint. If the alternative to intensified domestic production of maize and teff is importing
grain for domestic consumption at commercial prices, there is a high payoff to intensive maize and
teff production from society's perspective. Profits are high and stable even when world fertilizer
prices are high.

Whether intensified maize produced for export is profitable at the societal level depends on
the prevailing export price. Our analysis looks quite different if we assume that maize
produced with purchased inputs will be exported rather than serving as an import substitute.

In 1997, Ethiopia exported maize to Kenya at a price of $194/ton CIF Mombasa (considered to
be an unusually high price, influenced by a maize crop failure that year in Kenya). At this price
level low-input and intensive maize production remains profitable for maize farmers in Jimma and
West Shoa, although net gains are much lower compared to maize valued at import parity levels.

Future export prices are likely to be lower. The U.S. could supply maize CIF Mombasa at
$161/ton or less; in West Shoa, break-even export prices range from $133/ton (local seed) to
$151/ton (CIF Mombasa) for farmers using improved seed and fertilizer at or above
recommended levels. Break-even export prices for Jimma are even higher, ranging from $153-
$160/ton. This finding raises questions about whether Ethiopian maize will be able to compete
with lower-cost producers in the regional market.

Role of Transport Costs: Subsector marketing and transport costs for maize and inputs will have
to be reduced for Ethiopian maize to compete effectively on regional export markets. Transport
costs between the port and farmgate add 22-57% to the CIF cost of DAP and urea in West Shoa,
27-68% in Jimma, and 22-58% in Debre Zeit.

Accounting for the cost of extension and credit reduces the economic profitability of teff
significantly; the impact on maize depends on whether it is an import substitute or an
export. In the economic analysis, net income represents the residual return to factors that
facilitate crop production but are not explicitly costed out in the analysis. As a result of fertilizer
market liberalization the price of fertilizer is no longer directly subsidized by the government.
However, the government and donors are still facilitating farmer access to fertilizer in other ways,









i.e., through the provision of intensive extension assistance and credit programs. These program
costs are not recovered through the prices farmers currently pay for fertilizer.

If maize produced is considered to be an import substitute, the relative costs of extension and
credit provision are low and have little impact on the net gains to society. When maize is an
import substitute, accounting for program costs reduces net gains by only 11 to 15% in Jimma
and West Shoa. If maize is exported, however, deducting program costs reduces net gains of
exported maize by 39-60%.

As in the case of exported maize, accounting for the costs of extension and credit severely
reduces net gains from intensive teff production-by 42-890/-depending on assumptions about
fertilizer prices and levels of inputs used.


SUSTAINING THE MOMENTUM

Our results indicate that (1) the use of improved technology significantly increases yields and
income for participant farmers; and (2) as long as increased production can be consumed locally,
MOA/SG programs to promote these improved technologies are economically profitable. These
results support the Government of Ethiopia's (GOE) current policy of encouraging sustained use
of these new technologies by the early adopters and extending their use to a broader base of
farmers.

Nevertheless, the recent transition from the relatively small-scale MOA/SG program to the much
more ambitious goals of the NEP (almost 3,000,000 participants anticipated in 1998) reveals that
there are numerous challenges to meet if this momentum is to be sustained. Some of the more
important challenges are discussed below.

Expanding NEP Extension Coverage to A Wider Base of Farmers Without (1) Diluting the
Quality of the Extension Message, and (2) Increasing Program Costs

Issues ofScale: Increases in the number of DAs (extension agents) have not kept pace with the
rapid expansion in the numbers of NEP participants. Extension experts recommend a ratio of
about one DA per 100 demonstration plots, but the ratio now ranges from one DA per 150 to as
many as 500 demonstration plots in some areas.

Broadening the Scope of Coverage: The movement to less favorable agroecological zones and
poorer farmers suggests that yield response will diminish and the production risks will increase as
the program reaches out to a broader population of farmers. Agents will increasingly be working
with farmers who need more supervision than the previous round of participants, but will have
less time to devote to each one.

Cost Issues: To date, program expansion has been achieved largely through increases in the









number of demonstration plots supervised by individual agents, but there are limits to how far this
can go and legitimate concerns about the probable costs in terms of decreased quality of services.
The NEP's failure to "wean" farmers from the program after two years is another factor pushing
up costs as it limits the number of farmers trained and raises the costs per farmer.

Finding Low-cost, Financially Sound Ways of Administering Input Credit

Hidden but High Administrative Costs: The DAs, as well as other government personnel at the
regional, zone, and wereda levels, are heavily involved in the administration of the extension
credit. The current division of tasks appears to increase the public cost of the credit program
because many of the tasks performed by government personnel are ones that would normally be
performed by farmer organizations or banking sector personnel. Creative solutions are needed to
alleviate these pressures.

Competition between Extension and Regular Credit: Another problem is the strong competition
between the regular and the extension program for the limited, and sometimes declining, portfolio
of available credit. In several zones and weredas the entire credit allotment was used for the
NEP, leaving nothing for nonparticipants accustomed to getting credit through their cooperatives.
The underlying reason appears to be the limited amount of credit allocated by the regional
government, coupled with a desire to meet NEP targets. If this phenomenon continues, it may
seriously compromise the ability of NEP participants to become true graduates capable of
purchasing improved technologies on their own.

Reimbursement remains an important issue for the viability of the credit program. Although
overall rates remain acceptable, they are lower than those obtained by SG2000 and will probably
decline as the NEP moves to less favorable agroecological zones. Higher defaults will increase
government costs and reduce the economic profitability of the program. In an effort to minimize
their losses, regional governments have frequently called on the local police to enforce repayment.
This has resulted in some confiscation of farm assets to meet payments. While farmers must
reimburse if the credit system is to remain viable, liquidation of assets must be the exception
rather than the standard procedure. As the program moves to more risky farming situations, some
type of renegotiated payment schedules should be considered to allow for truly poor harvests.

Developing a Transparent and Responsive Input Supply System That Can Function
Independently of the NEP and Provide All Farmers Access to Low-cost, High-quality
Agricultural Inputs

Fertilizer Distribution: Efforts to decentralize decision making to the regional level have made it
difficult to fully liberalize fertilizer markets. In Amhara, Tigray, and southern regions the market
is dominated by a single firm owned in whole or in part by the regional government. These firms
are generally selected by the regional, zone, or wereda governments as the principal suppliers for
fertilizer made available through both the NEP and the regular credit programs. Given the role
that the government plays in allocating credit, it is easy for government agents to control which









firm receives the supply orders for fertilizer purchased on credit.


In a fourth region (Oromiya), a government-owned firm exists but the regional government opted
for a system of open bidding in 1998. The government firm competed with other distributors in a
bidding process that awarded supply contracts for fertilizer purchased with NEP and regular
credit to the lowest bidder, providing they could show proof of stocks required to fill the bid.
Preliminary results of price analysis studies suggest that the Oromiya bidding system provides
farmers with lower cost fertilizer (after controlling for transport and other cost factors) than the
systems in the other three regions.

Another key issue is the lack of differentiation of participants across functional areas. Importers
are performing the full range of import, wholesale and retail transactions, leaving very little of the
market for the large number of independent wholesalers and retailers who have been trained by
the NFIA in view of increasing competition in local markets. Further complicating the picture is
the fact that many service cooperatives tend to overestimate their members' demand for credit
purchases. This can result in substantial overstocks that the cooperatives try to liquidate through
cash sales, providing further competition for the independent retailers.

Policy Uncertainty Is A Problemfor Importers and Distributors: In 1998 fertilizer importers and
distributors faced considerable policy uncertainty: stated fertilizer polices were often abandoned
and markets were not as open as the fertilizer distributors originally perceived. The national
government has stated its goal of developing a free market for fertilizer, but regional policies
often carry different and changing messages. Policy uncertainties at both the national and regional
levels raise the cost of investing in the fertilizer sector and may discourage new entrants.

Reducing Costs: Input costs (especially fertilizer, but also seed) are a large component of the
financial budgets for farmers in the MOA/SG program. Reducing the cost of inputs will enhance
the accessibility of the program to a broader population and make the technology more profitable
for early adopters. Beyond increasing transparency and competition in the markets, there are a
number of other potential areas for cost reduction: (1) better timing of imports (importing earlier
to take advantage of seasonal drops in world prices and complete inland distribution before roads
deteriorate because of the rains; (2) larger allotments of foreign exchange to get greater
economies of size/scale in importing; (3) improved infrastructure (options mentioned were better
roads and the building of an inland storage facility just inside the Ethiopian boarder to reduce
excess port charges- these options would benefit other importers and exporters as well); (4)
helping farmers' organizations to take on more of the responsibilities for credit administration and
input delivery; and (5) better estimates of demand to avoid the expense of carry-over stocks.

Access to Improved Seed: Seed market development lags behind that of the fertilizer sector,
despite the urgent need to increase the availability of maize hybrids being promoted by the NEP.
Improved varieties of teff seed are available in local shops and markets, but hybrid maize seed is
not. Furthermore, there is no credit available for hybrid seed purchased outside of the NEP.
Apart from the National Seed Enterprise, a government parastatal that supplies the majority of









hybrid maize seed, the multinational firm Pioneer Hi-bred is the only other major actor in the
Ethiopian maize seed industry. Pioneer usually sells its hybrids to the National Seed Enterprise,
but is now beginning to promote direct cash sales to farmers. Both the credit and supply
constraints of hybrids have important implications for NEP graduates as without hybrid seed the
responsiveness of fertilizer declines dramatically.









TABLE OF CONTENTS

CONTRIBUTORS TO THE STUDY ......................................... iii

ACKNOWLEDGMENTS .................................................... v

EXECUTIVE SUMMARY .................................................. vii

LIST OF TABLES ..................................................... xvii

LIST OF FIGURES ..................................................... xviii

LIST OF ACRONYMS .................................................... xix

Section Page

1. INTRODUCTION ....................................................... 1

1.1. Objectives .................................... ................ 1
1.2. Methods ........................................................ 2
1.2.1. Sample Selection .......................................... 2
1.2.2. Questionnaire Design and Data Collection ........................ 5
1.3. Organization of the Paper ............................................ 6

2. CHARACTERISTICS OF PROGRAM PARTICIPANTS COMPARED TO TYPICAL
AGRICULTURAL HOUSEHOLDS ......................................... 7

3. YIELD RESULTS AND FACTORS AFFECTING CROP YIELDS AND TECHNOLOGY
CHOICE ............................................................. 10

3.1. Yield Results by Plot Type .......................................... 10
3.1.1. Maize ............................................. 10
3.1.2. Teff ................................. ............... 12
3.1.3. Moving From Analysis by Plot Type to Analysis by
Technology Type ......................................... 12
3.2. Factors Affecting Maize Yields .................................... 13
3.2.1. Types of Maize Technologies and Their Yields ................... 13
3.2.2. Econometric Analysis of Maize Yield Determinants ............... 15
3.2.3. Descriptive Analysis of Factors Related to Maize
Technology Choice ....................................... 18
3.2.4. Graduate Farmers' Decisions Concerning Choice of
Maize Technology ................ ...................... 19
3.3. Factors Affecting Teff Yields ..................................... 21
3.3.1. Teff Technology Types and Their Yields ........................ 21
3.3.2. Econometric Analysis of Teff Yield Determinants ................. 24









3.3.3. Graduate Farmers' Decisions Concerning Choice of Teff Technology .. 27

4. FINANCIAL ANALYSIS ................................................. 30

4.1. Data and Methods Used ............................................ 30
4.2. The Use of Improved Technology for Maize and TeffIs Extremely Profitable, Even if
Output Prices Decline by 25% or 50% .................................. 31
4.2.1. Jimma-Maize. .......................................... 32
4.2.2. West Shoa-Maize ....................................... 33
4.2.3. East Shoa- Teff ................................... .. . 33
4.3. Gains from Storage and Use of Storage Insecticide ......................... 35
4.3.1. Teff ................................................. 35
4.3.2. Maize ................................................. 35
4.4. Improved Seed and Fertilizer Costs Represent 50-75% of Total Costs .......... 35

5. ECONOMIC ANALYSIS ................................................. 39

5.1. Differences Between Financial and Economic Analysis ...................... 40
5.2. Method Used to Determine the Economic Values of Traded Items ............. 40
5.2.1. Parallel Exchange Rate .................................. 41
5.2.2. Import and Export Parity Prices ............................ 41
5.2.3. Estimating Program Costs ............................... 42
5.3. Economic Analysis: Summary of Main Findings ........................... 43
5.3.1. Economic Profitability of Intensive Maize and Teff Production ....... 44
5.3.2. Variable Profitability of Maize for Export ....................... 45
5.3.3. Extension and Credit Costs Reduce Profitability .................. 47

6. SUSTAINING THE MOMENTUM ....................................... 52

6.1. Expanding the NEP Extension Coverage to a Broader Group of Farmers ........ 52
6.1.1. Issues of Scale .......................................... 53
6.1.2. Broadening the Scope of Coverage ........................... 53
6.1.3. Cost Issues ............................................ 53
6.2. Improving the Credit System ........................................ 54
6.2.1. Evolution of the "Regular" Credit System ...................... 54
6.2.2. The Beginning of SG2000 and NEP Credit ..................... 55
6.3. Developing a Transparent, Responsive, Low-cost Input Supply System ......... 57
6.3.1. Encouraging Transparency and Competition in the Fertilizer Sector ... 57
6.3.2. Reducing Costs .......................................... 61
6.3.3. MeetingNeeds of All Farmers ................................ 61

REFEREN CES ................................. ....................... . 63









LIST OF TABLES


Table Page

Table 1. Sample Composition ...............................................4
Table 2. Agroecological Characteristics and Recommended Technology Packages ......... 4
Table 3. Characteristics of Participant Households and Average (CSA) Farm Households ... 8
Table 4. Fertilizer Use of Participants and Average (CSA) Farm Households ............. 9
Table 5. Yield Results by Zone, Plot Type, and Yield Tercile ...................... 11
Table 6. Types of Maize Technology Represented in the Sample ..................... 14
Table 7. Disaggregation of Maize Technology Types by Zone ...................... 15
Table 8. Regression Analysis of Factors Affecting Maize Yields in Jimma and West Shoa .. 17
Table 9. Labor Use in Maize Production by Technology Group ...................... 19
Table 10. Graduate Farmers' Response to Recommended Maize Practices ............... 20
Table 11. Reasons Given By Maize Graduates in West Shoa for Not Continuing A Given
Technology Component ............................................ 21
Table 12. Teff Seed and Fertilizer Use in East Shoa by Different Plot Types ............. 23
Table 13. Seeding Rate and Grain Yields by Teff Seed Varieties Used in East Shoa ....... 23
Table 14. Teff Technology Types Represented in the Sample ......................... 25
Table 15. Regression Analysis of Factors Affecting Teff Yields in East Shoa (Model 1) ..... 26
Table 16. Regression Analysis of Factors Affecting Teff Yields in East Shoa (Model 2) ..... 26
Table 17. Graduate Farmers' Response to Recommended Teff Practices, East Shoa ....... 28
Table 18. Most Common Reasons Given by Graduates in East Shoa for Not Continuing A Given
Teff Technology Component ........................................ 28
Table 19. Summary of Maize Results: Financial Analyses by Zone, Program Type and Input
Level .......................................................... 37
Table 20. Summary of Teff Results: Financial Analyses by Zone, Program Type and Input Leved8
Table 21. NEP Extension Intervention Budget for the 1995 Crop Season ................ 43
Table 22. Summary of Economic Analysis Results for Maize by Zone, Program Type and Input
Level .......................................................... 49
Table 23. Summary of Economic Analysis Results for Teff by Zone, Program Type and Input
Level ........................................................... 50
Table 24. Break-Even Export Prices for Maize ................................. 51
Table 25. Fertilizer Credit in the Three MOA/SG Study Zones, Oromiya Region .......... 56
Table 26. Fertilizer Import Trends: 1993 to 1998 ................................ 59









LIST OF FIGURES


Figure Page

Figure 1. Location of Survey Sites. .................. ......................... 3


xviii








LIST OF ACRONYMS


AFR/SD/PSGE

AISE
CERES
CSA
DA
DAP
EAL
EARO
EGTE
FA
FDRE
GMRP
GPS
MOA
MOA/SG

MSU
NEP
PLGY
SG or SG2000
SC
TGE
SAID

1 quintal = 100 kg


Africa Bureau's Office of Sustainable Development, Productive
Sector Growth and Environment Division
Agricultural Input Supply Enterprise
Crop-Environment Resource Synthesis
Central Statistical Authority, Ethiopian Government
Development Agent
Diammonium Phosphate
Ethiopia Amalgamated Ltd.
Ethiopian Agricultural Research Organization
Ethiopian Grain Trading Enterprise
Farmer Association
Federal Democratic Republic of Ethiopia
Grain Marketing Research Project
Global Positioning System
Ministry of Agriculture, Ethiopia
Ministry of Agriculture/Sasakawa Global 2000 Demonstration
Program
Michigan State University
New Extension Program, Government of Ethiopia
POLYGON program
Sasakawa Global 2000
Service Cooperative
Transitional Government of Ethiopia
United States Agency for International Development









1. INTRODUCTION


Ethiopia, one of the most densely populated countries in Africa, faces increasing food insecurity
unless it can dramatically boost agricultural productivity per hectare. In 1993, the
Sasakawa/Global 2000 Program (SG2000) began work in Ethiopia in partnership with the
Ministry of Agriculture's Department of Extension and Cooperatives (MOA). The objective of
their joint program was to demonstrate the productivity increases that could be achieved when
farmers were provided with appropriate extension messages, adequate extension contact, and
agricultural inputs such as improved seed, fertilizers and agrochemicals, delivered on time at
reasonable prices (SG2000 1996).

The MOA/SG2000 program provided participating farmers with improved inputs on credit in
amounts that were sufficient for one-half hectare demonstration plots. Farmers also received very
close supervision from extension agents during critical periods in the cropping cycle. Participants
agreed to provide land for the demonstration plot and to make a 25-50% down payment on the
input package before planting, with the balance due after harvest. In 1995, the MOA/SG2000
demonstration program reached more than 3,500 farmers in four regions. During the same year
the MOA launched the New Extension Program (NEP) funded and managed independently of the
MOA/SG program, but based on SG2000 principles. By 1997, the NEP was managing the bulk
of the demonstration plots (about 650,000) as the MOA/SG program reduced its direct
participation in the demonstration program to about 2,000 plots.

Although the MOA/SG program is widely considered to be a success, no formal analysis has been
carried out to determine the farm-level profitability of the program's improved technology
packages. In September 1997 MOA/SG agreed to collaborate with the Grain Marketing Research
Project (GMRP) and the Department of Agricultural Economics at Michigan State University
(MSU) to analyze the financial returns to the recommended technology packages and determine
the major factors affecting yield response.


1.1. Objectives

Our specific research objectives were to:

1. Describe the (a) characteristics of program participants versus nonparticipants; (b)
input use patterns on program plots versus selected types of nonprogram plots; and (c)
yield response by plot type and level of inputs used;

2. Evaluate the financial (i.e., private) and economic (i.e., social) profitability of selected
plot types and input levels, with particular attention to the profitability of the
recommended packages;








3. Analyze the relative contribution to yield of (a) different types of technologies, (b)
environmental factors, and (c) management practices; and

4. Describe the key challenges faced by the government in its effort to expand the
SG/NEP program, with particular attention to how the expansion is being affected by
government efforts to decentralize decision making and liberalize and privatize input
markets.


1.2. Methods

1.2.1. Sample Selection

The study examined the experiences of 1997 participants and graduates of the MOA/SG program
for maize and teffin three zones located in the Oromiya Region: (1) West Shoa (maize); (2)
Jimma (maize); and (3) East Shoa (teff). Figure 1 shows the location of the survey sites. Maize
and teff were chosen because they have been the major foci for the MOA/SG program. Within
each area, the study team (in consultation with MOA/SG staff) chose weredas that were
agroecologically homogenous and had a large number of current and graduate MOA/SG
participants (Table 1). All three zones are considered to have good to excellent agroecological
conditions for teff and/or maize production (Table 2).

In the case of teff, the study included 1997 participants in the MOA's NEP program because the
MOA/SG program focused on an experimental plant hormone to reduce lodging that had not yet
been extensively tested. The NEP technology package is the same one used by MOA/SG in
previous years: improved seeds, DAP and urea, and herbicides.

Within each zone, the team worked with local extension officials to construct a list of 1997 maize
and teff program participants and graduates (i.e., farmers who had previously participated in the
MOA/SG program, usually for two years). A total of 383 farmers were included in the sample.3
In each of the three zones, 40-80 current participants and 40-60 graduates were chosen. In
several cases all listed farmers were included in the sample. When it was necessary to make a
selection among farmers this was done randomly.

In East Shoa (teff) and Jimma (maize), current program participants usually had one or more plots
where they were using technology combinations different from the program plot. The survey
collected information about yield, area and input use for these "non-program" plots, in addition to
the program plots.





3Three households were subsequently dropped from the analysis because of missing yield data.










Table 1 summarizes the number of farmers in each category, and Table 2 describes key
agroecological characteristics of each zone and the recommended technology package.


Table 1. Sample Composition
CharacteristicO Current Traditional Plots MOA/SG
Study Sample Total Number MOA/SG of MOA/SG Program
Zone 0 Crop Weredas of Farmers Participants Participants Graduates

East Shoa Teff Ada 120 60 60 60


West Shoa Maize Woliso, 152 94 0 58
Wanchi

Jimma Maize Dedo, Kersa, 111 72 47 39
Seka
Chekorsa
Source: GMRP/MSU/SG2000/MOA 1997 Survey, Supervisor Field Reports.




Table 2. Agroecological Characteristics and Recommended Technology Packages
Altitude
Characteristics (meters Avg.
Study above sea Rainfall Recommended Inputs Per
Zone 0 Crop level) (mm) Soils Major Crops Half Hectare Plot

East Shoa Teff 1850 850 black, teff, wheat, 50 kg DAP,50 kg urea,
gray, red pulses 17.5 kg improved seed,
herbicide

West Shoa Maize 1600-2800 1420 red, gray teff, barley, 50 kg DAP, 50 kg urea,
maize, 12.5 kg BH660 hybrid
sorghum, maize seed, 80 cm
pulses between rows, 50 cm
between plants

Jimma Maize 1060-3000 1400- red teff, maize, Same as above
2000 wheat,
sorghum,
barley, pulses
Source: GMRP/MSU/SG2000/MOA 1997 Survey, Supervisor Field Reports.








1.2.2. Questionnaire Design and Data Collection


Primary data collection was carried out in two rounds between October and December 1997.
During these rounds the survey coordinator was assisted by three supervisors. Ten to fifteen
enumerators were hired in each zone. Training and questionnaire pre-testing were carried out
during the second and third weeks of October 1997. Primary data collection began the last week
of October and continued through mid-December 1997.

Additional data on the fertilizer subsector were collected in August 1998 to better understand the
characteristics of the rapidly evolving fertilizer subsector and estimate the costs of procuring and
transporting fertilizer to survey areas. Data were collected through interviews with private sector
importers, wholesalers, retailers, zonal and wereda-level agricultural officials, and a review of
secondary documents.

Yield Estimation and Area Measurement: The objectives of the first round of data collection
were to mark crop cut areas; harvest and weigh the grain from sample plots for yield estimation;
complete a short questionnaire on soil characteristics and history of the field being sampled;
measure field area; and geo-reference the field site using handheld global positioning system
(GPS) equipment (details on methods used for yield and area measurements and a copy of the
questionnaire used in the first round of the survey are available at
http://aec.msu.edu/agecon/fs2/papers/index.htm#recentidwp).

Collection of Demographic and Input Use Data: The objectives of the second round of data
collection were to gather demographic data on the household; general information for the whole
farm on area/input use for major crops and changes in livestock holdings over the past five years;
and specific information for the program, traditional, or graduate plot regarding (a) dates of major
field activities; (b) household and non-household labor inputs and costs; (c) amounts used and
costs of non-labor inputs (including animals, tractor, fertilizer, seed); (d) farmer perceptions of the
importance of purchased inputs; (e) farmer assessment of risk factors affecting maize/teff yield
during the past five years, including rainfall, hail/frost, wild animals, insects, plant diseases and
weeds; (f) farmer opinion of extension services received; and (g) marketing/consumption of
maize/teff over the last five years. (A copy of the questionnaire used for the second round of the
survey is available at http://aec.msu.edu/agecon/fs2/papers/index.htm#recentidwp.)

Data Entry, Cleaning, and Analysis: All primary survey data were entered by clerks at Addis
Ababa University during December 1997 and January 1998. Data cleaning and preliminary
analysis were carried out at Michigan State University in February 1998 by GMRP, Ethiopian
Agricultural Research Organization (EARO), and MSU researchers. Preliminary analyses
(financial budgets and yield determinant models) were completed in September 1998 and reported
in Howard et al. 1998. The current document is an expanded version of Howard et al. 1998. The
main addition is an economic (i.e., social) analysis of program profitability using a variety of
scenarios for import and export parity prices of cereals and fertilizer and a scenario that
incorporates government costs for extension and credit support to the program.










1.3. Organization of the Paper


Presentation of findings begin in Section 2 with a brief comparison of participant and non-
participant characteristics. Section 3 presents yields obtained by sample farmers and focuses on
factors affecting yield variability across plots and farmers' choice of technology. Section 4
summarizes farm-level profitability across the different program types and input levels. Section 5
adopts a broader perspective, presenting results of an economic (i.e., social) analysis of the
technologies that corrects for price distortions and government/donor costs of supporting
SG/NEP activities. Section 6 looks further into the off-farm aspects of the SG/NEP programs,
examining a number of difficult input and extension issues that must be dealt with as the SG/NEP
program expands.









2. CHARACTERISTICS OF PROGRAM PARTICIPANTS COMPARED TO TYPICAL
AGRICULTURAL HOUSEHOLDS


There is a tendency for extension programs to introduce new technologies to farmers with better
resources and skills first, expanding the program to others once the yield response and
profitability of the technologies have been demonstrated. To assess the extent to which this is the
case with the MOA/SG program in Ethiopia we have compared selected characteristics of
program participants covered by our 1997 survey with those of typical agricultural households
located in the same vicinity and covered by the CSA Agricultural Sample Surveys conducted in
1995/96.

The picture emerging from these comparisons is that the households in the survey of program
participants are better off than the average farm household (Table 3).4 Participant households
have 25 to 58% more people (i.e., more available labor); they cultivate 50 to 100% more land;
and there is 10 to 72% more land per capital. Participant households appear to be wealthier (more
livestock and traction animals), and are much more likely to have a literate household head than
the typical households described by the CSA data.

As expected, the proportion of the general farm population using improved seed and fertilizer is
quite low. The CSA Agricultural Sample Survey shows that less than 1% of farmers in the general
area of our 1997 survey used improved seed for any crop during 1995/96. Among CSA farmers in
the general area of West Shoa, only 4% used fertilizer on maize; the proportion increased to 27
for CSA farmers in the Jimma area. There is a longer history of fertilizer use on teff, however,
and 82% of CSA teff farmers in the area around East Shoa reported using fertilizer in 1995/96.

Also as expected, the highest fertilizer application rates are associated with the program plots of
current SG participants who used slightly greater than recommended doses. These doses
exceeded those used by graduate farmers and the general farm population represented in the CSA
survey (Table 4).5 What is of more interest, however, is to compare fertilizer doses on non-
program plots of current participants to regional and national averages, as this gives a hint as to
whether current participants were already ahead of their fertilizer-using neighbors before joining
the program. The non-program teff plots cultivated by participants averaged 147 kg of fertilizer
per hectare-substantially higher than the 110 kg average in the Oromiya Region and 99 kg





4As the CSA survey collected relatively few observations in any given wereda, the averages for the CSA
comparisons are based on households in the weredas covered by our 1997 survey plus weredas that were located
close to those in the survey (see full list in notes to Table 3).

5All three participant survey zones are located in the Oromiya Region; this is the most disaggregated level of
fertilizer use data we were able to get from CSA for 1996/97.









average nationally.6 The same pattern is apparent for maize, though the differences are not as
pronounced. Non-program plots cultivated by participants averaged 92 kg/ha, while fields
fertilized by farmers in the general CSA sample received 86 kg/ha in the Oromiya Region and 76
kg/ha nationally. Because the non-program plots of current participants represent the farming
practices used before using the MOA/SG technologies, these results suggest that participant
farmers were already using higher than average fertilizer doses before joining the MOA/SG
program.

We conclude from the above discussion that participants in the MOA/SG program have better
resource endowments than the general farm population and a tendency to use higher levels of
improved seed and fertilizer- even without program participation- than their neighbors. These
findings have implications for the expansion of the SG/NEP technologies to farmers who are more
resource constrained and less accustomed to using improved inputs than the current participants
(see Section 6 for further discussion).


Table 3. Characteristics of Participant Households and Average (CSA) Farm Households
East Shoa (teff)' West Shoa (maize)2 Jimma (maize)3 Ethi-
opia
Partici- CSA Partici- CSA Partici- CSA opia
pants Farmers pants Farmers pants Farmers
Mean area cultivated (ha/household) 3.0 2.0 2.6 1.5 2.1 1.0 1.0
Mean household size 7.1 5.7 8.7 5.5 7.4 5.0 5.2
(persons/household)
Mean hectares cultivated per capital4 0.62 0.36 0.34 0.31 0.31 0.21 0.21
Percent of literate household heads 95 22 85 36 95 19 22
Mean livestock units per household5 5.1 4.7 5.4 4.0 4.7 3.1 3.5
Mean number of draft animals per 2.7 1.9 2.3 1.7 2.3 1.5 1.1
household
Sources: GMRP/MSU/SG2000/MSU 1997 data; MSU analysis of CSA agricultural data base for meher crops,
1995/96.
'205 households from the CSA survey were used in the East Shoa analysis which covered Boset, Lome, Ada,
Dugda, Arsi Negele, Shashemene, Seraro, and Akaki weredas.
2221 households from the CSA survey were used for the West Shoa analysis which covered Woliso, Becho, Ambo
Zuria, Dano, Wonchi, and Dendi weredas.
'478 households from the CSA survey were used in the Jimma analysis which covered Limu Seka, Limu Kosa,
Sokoru, Tiro Afeta, Kersa Mana, Goma, Gera, Seka Cherkorsa, Dedo, and Omanada weredas.
4 Calculated at the household level first, then averaged across households to give each household equal weight in
the calculation; note that the same result will not be obtained when dividing sample mean area by sample mean
population.
Calculated using following weights: cattle=l, sheep/goats =.5, horses/mules = .7.




6Note that the CSA data are application rates for fertilized fields only; fields receiving no fertilizer are not included
in the analysis.










Table 4. Fertilizer Use of Participants and Average (CSA) Farm Households
Type of Participant Plot CSA Farmers
Non- Fertilized
Program program Fields Only

(average kilograms of fertilizer product per
hectare)
Teff
East Shoa (SG/NEP only) 202 147 155
Oromiya Region (CSA only) 110
National (CSA only) 99

Maize
West Shoa (SG/NEP only) 216 -84
Jimma (SG/NEP only) 205 92 192
Oromiya Region (CSA only) 86
National (CSA only) 76
Source: GMRP/MSU/SG2000/MOA 1997 Survey; CSA Statistical Bulletin 171, 1996/97.









3. YIELD RESULTS AND FACTORS AFFECTING CROP YIELDS AND
TECHNOLOGY CHOICE


We now turn to the presentation and analysis of yield results from our sample farmers. In the
next section (3.1) we begin by examining the yield results obtained in the three plot types by zone
and crop: program participants' plots using the recommended technology package, traditional
plots grown by program participants, and plots managed by graduates using the technology of
their choice. In Sections 3.2 and 3.3 we turn to an examination of the key factors affecting maize
and teff yields and quantify the relative yield impact of these factors.


3.1. Yield Results by Plot Type

Average maize and teff yields across all survey plots far exceeded national and regional averages.
Yields for plots where farmers used high-input technologies were much greater than yields for
plots using low-input technologies. Greater heterogeneity than anticipated within given plot types
(due to variation in types and levels of inputs) produced mixed yield performance across the three
plot types. Table 5 summarizes these results, presenting yields by crop, study zone, and plot type.
The budgets also break each plot-type group into terciles (by crop yield) in order to examine the
variation within each plot type.


3.1.1. Maize

Jimma: Average maize yields on program and graduate plots in Jimma were 5.5 and 6.8 tons/ha,
respectively. These yields were more than double the 1996/97 national and Oromiya Region
averages (1.9 and 2.1 tons/ha, respectively). Program graduates in Jimma achieved yields that
were 1.3 tons/ha higher than program participants although they used essentially the same
technology (improved seed and approximately the same amount or less of DAP and urea). This
suggests the existence of a "learning curve" with farmers becoming more proficient in the
application of improved technology and management techniques over time.

Traditional plots in Jimma were generally cultivated with local seed and DAP. The mean yield for
these plots was 2.8 tons/ha. The large yield differences (more than 2 tons) between the traditional
plots and those cultivated with the improved technologies used on the program and graduate plots
clearly demonstrate the role improved inputs play in augmenting maize yields.

West Shoa: Program participants in this zone obtained slightly higher average yields than
participants in Jimma-5.6 tons/ha. The graduates did not do as well as those in Jimma, however.
West Shoa graduates were split into two groups: more than half reverted to traditional methods
(local seed without fertilizer) while the rest continued to use improved seed and recommended
levels of fertilizer. Those using low-input techniques achieved average yields of














Table 5. Yield Results by Zone, Plot Type, and Yield Tercile

(a) MOA/SG/NEP Program
Plot (b) Traditional Plot (c) Graduate Plot
Yield Tercile
Commodity/ Mean/ Mean/ Mean/
Zone Item 1 2 3 Total 1 2 3 Total 1 2 3 Total
MAIZE/ YIELD (ton/ha) 3.9 5.5 7.2 5.6* na na na na 2.9 4.6 6.9 4.8*
WESTSHOA nusedincalculations 30 31 31 92 na na na na 19 19 19 57

MAIZE/ YIELD (ton/ha) 4.1 5.4 6.9 5.5** 1.6 2.7 4.1 2.8** 5.1 6.9 8.4 6.8**
JIMMA n used in calculations 22 24 23 69 15 16 16 47 13 13 13 39

EAST GRAIN YIELD (ton/ha) 0.9 1.3 2 1.4 0.8 1.4 1.9 1.4 1.0 1.4 2.0 1.5
SHOA/Teff STRAW YIELD (ton/ha) 2.2 2.1 2.3 2.2 2.2 1.9 2.0 2.0 2.4 2.0 1.8 2.1
n used in calculations 20 20 20 60 20 20 20 60 21 18 21 60
Source: Field data from 1997 GMRP/MSU/AAU/SG2000/MOA Survey.
na = not applicable
* yield differences between MOA/SG and graduate plots were significant at the 95% level.
** yield differences between MOA/SG and traditional plots belonging to the same household were significant at the 95% level; yield differences between
MOA/SG and traditional plots, and MOA/SG and graduate plots, were also significant at the 95% level.











3.8 tons-far exceeding national and regional averages but still significantly lower than the 5.8
ton/ha yields obtained by those using the improved technologies.


3.1.2. Teff

Average grain yields for teffwere similar on all plots (program, graduate, and traditional)- about
1.4-1.5 tons/ha or 50-55% higher than national and regional average yields of 0.9 ton/ha and 1
ton/ha, respectively (FDRE 1997). The lack of a significant difference across plot types is due in
large part to the use of both improved seed and fertilizers on both traditional and graduate plots.
Further complicating the comparison is the flexibility of the 1997 teff program- participants
were allowed to partially adopt the recommended package by using different levels of fertilizer
(often less than recommended rates of urea) or substituting a different variety of seed.' Although
farmers often referred to these seeds as "local" varieties, according to Ethiopian Agricultural
Research Organization (EARO) researchers, in the East Shoa region these are more likely to be
saved seed from improved varieties that were previously distributed (e.g., DZ-Cross-37 or DZ-
01-196) than traditional teff varieties.

Production of teff straw, which is becoming an important commercial crop, ranged from 2 to 2.2
tons/ha across survey plots. We are not aware of any national or regional statistics on teff straw
yields that can be used for comparative purposes.

In summary, unlike the maize areas, most of the teff farmers surveyed had previously adopted the
key components of the improved technology package for use on all teff plots. The use of
improved technology allowed farmers to achieve yields that were substantially higher than
national and regional averages.


3.1.3. Moving From Analysis by Plot Type to Analysis by Technology Type

In the preceding paragraphs the yield results were examined for the three plot types by zone and
crop. For maize, the average yields obtained on program and graduate plots were significantly
higher than those obtained on traditional plots. However, there was a wide variation in the crop
yields within a given plot type, implying that the type and level of inputs used on these plots,
especially on the traditional and graduate plots, were not always homogeneous for a particular
crop and zone. For example, some maize graduates in West Shoa reverted to traditional
production methods (local seed and no fertilizer) while others opted to continue using inputs



7 This type of substitution was not permitted by the MOA/SG program but was introduced as responsibility for
demonstration plots shifted from MOA/SG to NEP.










similar to the recommended package; among the traditional teff plots, some farmers used
improved seed, DAP and urea while others used only improved seed and DAP.

The objective of the analysis in Sections 3.2 and 3.3 is to identify the key factors affecting maize
and teff yields and to quantify the relative yield impact of these factors. To accomplish this the
plots were grouped by types and levels of seed and fertilizer used rather than by the original
sampling criteria (type of farmer and plot). This permits better control for the various
technologies when examining the influence of other factors. The other factors examined fell into
two broad categories: (1) exogenous factors that farmers respond to but cannot completely
control such as rainfall, soil types, disease and pest attacks, and (2) endogenous factors linked to
management practices such as timing of critical operations, amount of labor used, and number of
plowings.

We proceed by (1) describing the principal technologies used and their yields, (2) presenting
econometric results that identify and quantify key yield determinants, (3) discussing descriptive
statistics for factors that are correlated with the use of improved technologies and, therefore,
thought to encourage their adoption, and (4) describing graduate farmers' decisions about
continued use of the high-input technologies. Section 3.2 discusses these topics for maize and
Section 3.3 does so for teff.


3.2. Factors Affecting Maize Yields

3.2.1. Types of Maize Technologies and Their Yields

We grouped maize plots into the four technology types described in Table 6. We consider
technology types 1 and 2 as 'low-input' technologies and types 3 and 4 as 'high-input'
technologies. The only difference between the two high-input technologies is the level of fertilizer
applied. In both high-input groups the amount of DAP and urea applied are equal within a given
plot but not across plots. For plots in technology type 3 application rates range from 50 to 98 kg
of each product per hectare; for plots in technology type 4 the rates range from the recommended
level of 100 kg per hectare to 208 kg. The median for type 4 plots is 113 kg/ha, hence at least
50% of the farmers in this group are applying fertilizer at approximately the recommended level.











Table 6. Types of Maize Technology Represented in the Sample
Number of Average Fertilizer
Plots Using a Average Applied (kg/ha)
Given Yields
Type of Maize Technology Technology (kg/ha) DAP UREA

(1) Local seed, no fertilizer 37 3639 0 0
(2) Local seed plus DAP 44 2918 103 0
(3) Improved seed plus DAP and urea at < 103 5910 86 86
recommended dose
(4) Improved seed plus DAP and urea at 118 5786 119 119
>= recommended dose
Source: Calculated from GMRP/MSU/SG2000/MOA survey data.


A comparison of yields for each technology group shows statistically significant differences
among all the groups but technology types 3 and 4. The lack of significant difference between
these two high-input groups is not unexpected as the difference in fertilizer application rates is not
very large. Surprisingly, however, the lowest technology level (local seed, no fertilizer)
performed better than the next higher technology (local seed plus DAP). This result in the
overall data is entirely attributable to 33 plots in the West Shoa zone cultivated by graduate
farmers, as illustrated by the zone-disaggregated data presented in Table 7.

Although there is substantial variability in the yields obtained by this group of farmers (1.9 to 6.8
tons/ha), the overall average was about 3.9 tons/ha-more than double the yields for the
comparable technology in Jimma (1.8 tons, but only 4 observations) and statistically higher than
the yields for the next higher technology in both Jimma (2.9 tons based on 43 observations) and
West Shoa (3.5 tons, but only 1 observation). Numerous checks were conducted on the data to
verify that these results are not due to measurement errors. Numerous cross-checks were also
conducted using both quantitative and qualitative variables in a search for clues as to why this
low-input technology performed so well in West Shoa during 1997.

No evidence was found of measurement error in the yield estimates, or clear explanation for the
good performance of this group of graduates. The leading hypotheses are: (1) the local variety
performs better under poor rainfall conditions (31 of the 34 farmers indicated that the total
quantity and/or the distribution of rainfall was poor in their village during the 1997 season), and
(2) there may have been some residual fertilizer effect from the previous year because most (25 of
34) of these farmers applied recommended levels of DAP and urea on the same plots in 1995/96.
Other possible mitigating factors might be differences in land quality. Farmers in this group were











more likely than those in other groups to report level fields (vs. gullied or sloped fields) and more
likely to report high fertility soils.


Table 7. Disaggregation of Maize Technology Types by Zone
Jimma West Shoa

Number Average Average kg/ha Number Average Average kg/ha
Technology Type of Yield of Plotsb Yield
Plotsa kg/ha DAP UREA kg/ha DAP UREA

(1) Local seed with no 4 1835 0 0 33 3858 0 0
fertilizer
(2) Local seed plus DAP 43 2905 100 0 1 3480 208 0
(3) Improved seed plus 58 6007 87 87 45 5784 86 86
DAP and urea at <
recommended dose

(4) Improved seed plus 50 5922 116 116 68 5685 121 121
DAP and urea at >-
recommended dose
Source: Calculated from GMRP/MSU/SG2000/MOA survey data.
a For Jimma, all the plots for technologies 1 and 2 were traditional plots; among the 58 type 3 plots, 24 were
graduates and 34 SG participants; among the 50 type 4 plots 15 were graduates and 35 SG participants.
b For West Shoa, the type 1 and type 2 plots were all graduate plots; among the 45 type 3 plots were 11 graduates
and 34 SG participants; among the 68 type 4 plots there were 10 graduates and 58 SG participants. Two farmers
in West Shoa using intermediate technologies (local seed plus DAP and urea and improved seed with no fertilizer)
were excluded from the analysis.


3.2.2. Econometric Analysis ofMaize Yield Determinants

The objective was to identify and evaluate the relative importance of different factors affecting
maize yields. It was planned to develop a model with a disaggregated set of input, crop
management, and environmental variables that combined observations from all the available
technology types. Unfortunately, the variability in the data did not allow this. The main problem
was very high correlation, among the input variables in particular, but also among some of the
other variables. For example, local seed was used exclusively with the two low-input
technologies while improved seed is used exclusively with both DAP and urea. This made it
impossible to separate the seed effect from the fertilizer effect. The application of DAP and urea
in equal quantities for all the high-input plots further increased problems of multicollinearity,











making it impossible to evaluate the relative contributions of urea and DAP. Similar problems
were encountered with some of the environmental and management variables as some tended to
be highly correlated with each other and frequently correlated with the dummy variable used to
differentiate the two maize production zones.

The data did, however, permit us to model the yield impact of the different technology types, and
some environmental factors and key management practices. Table 8 summarizes the results by
zone.

Contribution of Technology Factors on Maize Yields: Both in West Shoa and Jimma the high-
input technologies explain a large amount of the yield variation observed across plots. In Jimma
the technology types 3 and 4 (improved seed with both DAP and urea) yielded almost 3.5 tons/ha
more than the technology type 1 using local seed without fertilizer. In West Shoa, the high-input
technologies (type 3 and 4) also performed better than the low-input ones, but the difference was
substantially less than in Jimma-about 1.8 tons versus 3.5 tons increase when moving from low-
to high-input technologies. This is due primarily to the very high yields obtained by the graduate
farmers who opted to use local seed without fertilizer (see discussion in Section 3.1.1 above).

Contribution of Environmental Factors on Maize Yields: An important environmental factor
contributing to the maize yield variability in Jimma was soil type-red soil produced almost 1.5
tons more output than gray or black soils in this zone. Fortunately 95% of farmers planted maize
on red soils. Several soil variables were also significant in explaining the yield variability in West
Shoa. Farmers in West Shoa who declared that their soils were poor (only 20% of the West Shoa
sample) obtained yields about 700 kg lower than farmers having declared soil of medium or high
fertility. Red soil was associated with yield increases of about 400 tons per hectare. But unlike
Jimma, red soils did not significantly impact the yield variability. About 87% of West Shoa
farmers planted maize on red soils.

Contribution of Management Practices on Maize Yields: Certain management practices also
affect yields. In Jimma, yields declined by about 200 kg for each week a farmer deviated from the
optimal planting week. The average deviation was 1.3 weeks, suggesting an average loss due to
late planting of 260 kg/ha. Each centimeter of deviation from recommended row distances
resulted is a yield loss of about 35 kg/ha. In Jimma, the average deviation was 9 centimeters,
resulting in an average yield loss due to poor row spacing of 315 kg/ha. The last management
variable in the Jimma model that appears to be related to yields (.07 level of significance) is the
number of plowings. Farmers in the Jimma zone who plowed their fields more than four times
before planting got about 550 kg more per hectare than those who plowed fewer times-84%
plowed at least four times so this fairly substantial yield loss affected only 16% of farmers.











Table 8. Regression Analysis of Factors Affecting Maize Yields in Jimma and West Shoa
Jimma (Adj R2 0.55) West Shoa (Adj R2 0.30)

Variables Coef. SE Coef. T Sig T Coef. SECoef. T Sig T

Constant 1021 895 1.14 0.26 3723 588 6.33 .00

Tech2 dummy: 855 725 1.18 0.24 Only one case of this technology in
Local seed plus DAP the zone; omitted from model.

Tech3 dummy: 3459 727 4.76 0.00 1843 350 5.27 .00
Improved seed, lower than
recommended fertilizer use

Tech4 dummy: 3532 731 4.83 0.00 1803 317 5.68 .00
Improved seed,
recommended or higher
fertilizer use

Diffrowdistance: -35 14 -2.57 0.01 -29 17 -1.64 .10
Absolute deviation in cm.
from ideal distance

Diffplantdistance Not a significant variable in this -17 9 -1.79 .08
Absolute deviation (cm) zone.
from ideal plant distances

Diffplantdate: -194 90 -2.15 0.03 Not a significant variable in this
Absolute deviation (weeks) zone.
from ideal planting date

Diffweeddate Not a significant variable in this -124 79 -1.58 .12
Absolute deviation (weeks) zone.
from ideal weeding date

Plowing dummy 563 309 1.82 0.07 Not a significant variable in this
1 represents >4 plowings zone.

Red soil dummy 1448 533 2.71 0.01 432 372 1.16 .24
1-redsoil

Soil fertility assessment Not a significant variable in this 744 314 2.37 .02
dummy zone.
1= medium to high fertility
Source: Estimated from GMRP/MSU/SG2000/MOA survey data.











In West Shoa, the failure to follow recommendations concerning planting distances (both between
plants and between rows) reduced yields by about 15-30 kg per centimeter (Table 8). The average
deviation from the recommendation was 18 centimeters for plant spacing and 10 centimeters for
row spacing, resulting in an average loss of 280 kg/ha for planting distance errors and 300 kg/ha
for row spacing errors. Failure to weed at the optimal number of weeks following planting
reduced yields by about 125 kg per week. The average deviation from recommended weeding
time was 1.6 weeks, resulting in an average loss of 200 kg/ha.


3.2.3. Descriptive Analysis of Factors Related to Maize Technology Choice

These regression results present a rough picture of the factors that appear to have had the most
important influence on maize yields during the 1997 production season, but it is important to note
the models still explain only a portion of the variability in the yield data-55% for Jimma and 30%
for West Shoa.

Data was collected on a much wider range of variables representing both agroecological variables
and management practices. Attempts to incorporate a wider range of these variables in the models
were thwarted by a substantial amount of correlation among the variables and a fairly high degree
of compliance with recommended practices. This meant that there was not much variation in the
data for some variables of interest.

The set of variables concerning labor inputs is a case in point. These variables were not significant
in the multivariate yield models estimated, but there appeared to be some important differences in
labor use when households are classified by technology type. Table 9 summarizes the results of
bivariate analyses conducted to test differences in labor use by type of technology. Farmers in the
low-input technology groups used significantly less labor per hectare than those in high-input
groups. Interestingly, the major differences seemed to come from the access of the high-input
farmers to mutual and hired labor, as there is no statistically significant difference in the amount of
family labor used per hectare. This suggests that use of high-input technologies requires not only
more resources for the purchase of inputs but also more resources to attract non-family labor.











Table 9. Labor Use in Maize Production by Technology Group
Low-input High-input Mean Difference between Low-
Technology Technology input and High-input Technology
Type of Labor (Types 1 and 2) (Types 3 and 4) (Level of Significance)

Total labor days/hectare (8 87 122 .00
hrs/day)
Total weeding days/hectare 28 46 .00
Number of plowings 4.8 5.2 .00
Days of mutual labor/hectare 14 41 .00
Days of hired labor/hectare 9 14 .04
Days of family labor/hectare 65 66 .78
Source: Calculated from GMRP/MSU/SG2000/MOA survey data.


3.2.4. Graduate Farmers' Decisions Concerning Choice of Maize Technology

Graduate farmers in Jimma appear to be convinced that the recommended maize technologies and
production practices are worthwhile as 100% of the graduates surveyed in the zone are continuing
to use the new technologies (under the NEP program). This is not the case in West Shoa, where
60% of the graduate farmers in this zone discontinued the use of improved technologies in 1997.
Table 10 summarizes the behavior of graduates in both zones with respect to a number of key
recommendations.

It is noteworthy that the 60% of West Shoa farmers who abandoned the technology abandoned it
in its entirety, dropping both the improved seed and all fertilizers. The reasons for discontinuing
the technology package in West Shoa throw some light on the agronomics and the economics of
maize production in the zone (see Table 11). The predominant reason for discontinuing both the
seed and fertilizer technology was the unsatisfactory yield response in the previous year-a year
of relatively good rains but more hail damage than usual. Many farmers simply did not see
enough difference in yield to justify the added risk and expense of improved seed (hybrids) and
fertilizer. Some noted that they had suffered major crop losses due to hail and animal damage the
previous year but were still obliged to repay the credit. Two such farmers mentioned having sold
their oxen to take care of the debt. Another farmer claimed he was forced to sell his maize early
at very low prices in order to reimburse the credit-something he did not want to be forced into
doing again.











Table 10. Graduate Farmers' ResDonse to Recommended Maize Practices


Source: Calculated from GMRP/MSU/SG2000/MOA survey data.


Another frequently used explanation for abandoning the technology in West Shoa was the delayed
start of the rains in 1997, which substantially increased the risk of not covering input costs.
Farmers never specifically said that they believed local varieties performed better in years of poor
rainfall; most alluded to the desire to reduce the risk of not being able to reimburse credit and the
fact that the late planting could be done more quickly by broadcasting local varieties than by
planting improved varieties in rows. In general, farmers' responses suggest that risk aversion is
compelling them to dis-adopt improved seed and fertilizer technologies. Another frequently
mentioned constraint was the high cost of the inputs, particularly fertilizer, and an inability to
mobilize enough cash for the down payment. A less common explanation (two cases) was a land
constraint-i.e., the farmer could not find another farmer willing to combine land resources to
meet the minimum plot size of 0.5 hectares.8 In sum, given the relatively poor rains in the 1997
season, many farmers felt more secure using traditional maize technologies which are much less
expensive and, therefore, less risky.





SNote that the minimum size plot is only a requirement for participation in the demonstration plot program but
the only way MOA/SG 2000 graduates could get access to maize improved seed was to participate in the NEP
program -hence the mention of a land constraint.


West Shoa Jimma

Number of GraduatesO 58 39

Technology Component (percent continuing a given technology)

Improved seed 36 100

Recommended seeding rate 38 100

Planting in rows 38 100

Recommended rate of DAP 38 100

Recommended rate of urea 36 100

Recommended row spacing 36 100

Recommended spacing between plants 38 100












Table 11. Reasons Given by Maize Graduates in West Shoa for Not Continuing A Given
Technology Component
% of Graduates Not Using
Common Reasons for Discontinuing A Given Technology the Recommended Practice
Technology Component Component for A Given Reason

Improved seeds Unsatisfactory yield from previous year 54
Returned improved seed to distributor due to delayed rains 24
Inability to pay for the whole package 14
Seeding rate Broadcast seeding is easier and requires less labor 72

Row planting Demands fertilizer which is unaffordable 39
Requires more labor and time 28
Believes that broadcast is better 11

DAP rate Applying fertilizer has not increased yield 38
High cost of fertilizer 28
Shortage of rain 17
Urea rate High cost of fertilizer 27
Applying fertilizer has not increased yield 24
Shortage of rain 16

Row spacing No reasons given, but local seed is often broadcasted and
considered a more rapid method of planting in years when
the rains are late

Plant spacing As above
Source: Calculated from GRMP/MSU/SG2000/MOA survey data.


However, the follow-up report of the field supervisors in the West Shoa zone gives an impression
that a majority of graduates who had returned to traditional practices in 1997 for a variety of
reasons had re-adopted the technology package in the following year. Perhaps the strong
evidence of good yields enjoyed by other graduates who had continued the technology package in
1997 made many graduates realize the benefits of hybrid maize seeds and fertilizer even in poor
rain conditions.



3.3. Factors Affecting Teff Yields

3.3.1. Teff Technology Types and Their Yields

Compared to maize, the use of improved teff technology was much more widespread across the
program, traditional and graduate plots surveyed in the East Shoa region. All 180 farmers
surveyed (60 from each plot type) reported using improved seeds and only 6 said they used no










fertilizer at all. Despite this uniformity in the use of improved seeds and fertilizer across all plot
types, teff yields varied greatly, ranging from about half a ton to 3.3 tons per hectare.

The variation in the yield levels compelled us to look further into "the recommended teff
technology package" and the alterations in the package made by the farmers on their traditional
and graduate plots. As noted earlier, the teff technology package recommended by the extension
service consisted of 35 kg/ha improved seeds (Qoladma or Magna), 100 kg/ha each of DAP and
urea, and herbicides. However, the only mandatory requirement under the NEP program was the
purchase of equal (unspecified) quantities of DAP and urea. Unlike the maize program
participants, teff program farmers in East Shoa were allowed to use their own improved seeds
saved from previous harvests (in place of the distributed seeds, Qoladma and Magna) and any
quantity of herbicides. As indicated in Table 12 almost 50% of the current participants used their
own improved seeds retained from previous harvests. Also, unlike the maize technology package,
the seed component of the teff technology package was not exclusively for the current
participants. Qoladma and Magna were also available for use on graduate and participants'
traditional plots. Thus, nine current participants and three graduate farmers reported using the
improved seeds Qoladma or Magna on their traditional and graduate plots.

Because of the flexibility in the seed input, the teff technology used by the current participants
differed from the traditional and graduate plots only in the rate and ratio of DAP and urea
application and the seeding rate (Table 12). Farmers altered the recommended fertilizer
application rate by either lowering both the DAP and urea application rate, only the urea
application rate, or not applying urea at all. The seeding rate on traditional and graduate plots
was significantly higher than the seeding rate on participant plots. This is explained entirely by the
higher seeding rate used with farmer-saved seed (Table 13), which was used by a majority of
farmers on the graduate and traditional plots.

The seeding rate of farmer-saved seed was significantly higher than the seeding rates of Qoladma
and Magna, which is closer to the recommended seeding rate of 35 kg/ha. Yields from farmer-
saved seed were also significantly higher than the improved varieties Qoladma and Magna
distributed by the NEP program (Table 13).











Table 12. Teff Seed and Fertilizer Use in East Shoa by Different Plot Types
Technology Package Participant Plot Traditional Plot Graduate Plots All
Seed Number of Farmers Using
Qoladma 24 4 2 30
Magna 7 5 1 13
"Local" 29 51 57 137

Mean Application Rate (kg/ha)
Seed 47 (a,b) 57 (b,c) 69 (a,c) 58
DAP 104 89 95 96

Urea 104 58 59 72

Herbicide 0.45 0.27 0.36 0.36
Source: Calculated from GRMP/MSU/SG2000/MOA survey data.
a Differences in the seed application rate were significantly different at the 99% level.
b Differences in the seed application rate were significantly different at the 90% level.
c Differences in the seed application rate were significantly different at the 95% level.



Table 13. Seeding Rate and Grain Yields by Teff Seed Varieties Used in East Shoa
Seed Variety Seeding Rate (kg/ha)a Grain Yields (kg/ha)

Qoladma (N=30) 39 (a) 1096 (a)

Magna (N=13) 41 (b) 1110 (a)

Farmer saved (N=137) 63 (a,b) 1498 (a)

All 58 1403
Source: Calculated from GRMP/MSU/SG2000/MOA survey data.
aDifferences were statistically significantly at the 99% level.
bDifferences were statistically significant at the 95% level.


The teff plots are grouped into 5 technology types (described in Table 14) based on the various
combinations of improved seed type and fertilizer application rates found in the sample.
Technology type 1 corresponds to the full technology package recommended by the extension
service (100 kg DAP, 100 kg urea, Qoladma/Magna seed). Technology type 2 users apply near-
recommended quantities of fertilizer but use saved (improved) seed. Plots classified as
technology types 1 or 2 were generally those of current program participants. Those in










technology types 3 and 4 were traditional and graduate plots. In technology types 3 and 4, DAP
is applied at close to the recommended rate, but only 50% of the recommended rate of urea is
used. Technology type 5 is characterized by the use of improved seeds (mostly farmer-retained)
and no fertilizer. There was no significant variation in the herbicide application rate within a given
technology type. A majority of the farmers used some herbicide in all the technology groups, as
indicated in Table 14. In technology type 1 (full NEP package), herbicide was used by all 35
farmers.

A comparison of teff grain yields shows statistically significant differences between technology
types that use different seed varieties, but not between technology types with different urea rates.
Thus, yields for technology type 1 (Qoladma/Magna seed with recommended quantities of DAP
and urea) are significantly lower than technology type 2 (farmer-retained improved seed with
near-recommended DAP and urea). Similarly, yields for technology type 3 (Qoladma/Magna with
recommended quantity of DAP, 50% urea) are statistically lower than yields for technology type 4
(farmer-retained improved seed, recommended DAP, 50% urea). However, yield differences
between technology types 1 and 3 (Qoladma/Magna, recommended DAP, different urea rates)
and types 2 and 4 (farmer-retained improved seed, recommended DAP, different urea rates) are
not significant. Technology type 5 grain yields were not statistically different from yields of other
technology types.


3.3.2. Econometric Analysis of Teff Yield Determinants

Several models were developed to disaggregate the impacts of technology, environment, and crop
management variables on teff grain yields based on the 160 observations across all technology
types described in Table 14. Results from two of these models are summarized in Tables 15 and
16.

Contribution of Technology Factors to Teff Yields: The results from Model 1 indicate that
farmers who used saved seed (improved) (Groups 2 and 4) obtained yields that were about 300
kg/ha higher than yields when farmers used the seed distributed by NEP, Qoladma/Magna (Group
1). Reducing urea rates by half had no significant impact on teffyield-indicated by the
coefficients (and their lower significance level) of variables Technology 3 and Technology 5.
Herbicide application had a negative but statistically insignificant impact on yield variability.












Table 14. Teff Technology Types Represented in the Sample
Number of Average Average kg/ha of % of Farmers in a
Plots Using Yields Fertilizer Given Technology
a Given kg/ha Group Applying
Type of Teff Technology Technology DAP Urea Herbicides

(1) Improved seed (Qoladama or 35 1082 105 105 100
Magna) plus recommended
quantities of DAP and urea

(2) Farmer-retained improved 63 1523 96 96 73
seed plus near-recommended
quantities of DAP and urea

(3) Improved seed (Qoladma or 8 1181 95 47 88
Magna) plus near-recommended
quantities of DAP, 50% of
recommended urea

(4) Farmer-retained improved 69 1482 98 45 73
seed plus near-recommended
DAP, 50% of recommended
urea

(5) Any improved seed and no 5 1385 0 0 60
fertilizer
Source: Calculated from GMRP/MSU/SG2000/MOA Survey data.


Results from Model 2 (Table 16) also confirm the association between use of farmer-retained
improved seed and higher yields. Ceteris paribus, yields on plots where farmer retained seed was
used were about 300 kg/ha higher than plots where farmers used Qoladma or Magna seed.
Fortunately three-quarters of farmers in the sample used their own saved seeds.

Including kilograms per hectare of DAP and urea as variables in Model 2 did not change the
results very much: neither fertilizer nor herbicide use were significant factors explaining yield
variation.












Table 15. Regression Analysis of Factors Affecting Teff Yields in East Shoa (Model 1)
Variables Coef. SE Coef. T Sig T

Constant 1234 95 12.95 0.00

Tech2 dummy: Farmer-retained improved seed, near- 323 94 3.42 0.00
recommended quantities of DAP and urea

Tech3 dummy: Improved seed (Qoladma or Magna), 59 169 0.35 0.73
near recommended DAP, 50% urea

Tech4 dummy: Farmer-retained improved seed, near- 295 93 3.08 0.00
recommended DAP, 50% urea

Tech5 dummy: Any improved seed and no fertilizer 118 207 0.57 0.57

Herbicide (kg/ha) -43 103 -0.41 0.68

Plowing dummy:1 = more than 4 plowings -135 75 -1.81 0.07

Rain distribution dummy (farmer assessment): 344 92 3.74 0.00
1 = good/excellent distribution

Gray soil dummy:1= gray color soil -227 72 -3.14 0.00
Source: Calculated from GMRP/MSU/SG2000/MOA Survey data.
Note: Adjusted R2 = 0.24




Table 16. Regression Analysis of Factors Affecting Teff Yields in East Shoa (Model 2)
Variables Coef. SE Coef. T Sig T

Constant 1227 105 11.7 0.00

Seed type dummy:1 = farmer retained improved seed 295 82 3.6 0.00

DAP (kg/ha) -0.17 0.8 -0.2 0.84

Urea (kg/ha) 0.33 0.9 0.4 0.71

Herbicide (kg/ha) -43 107 -0.4 0.68

Plowing dummy: = more than 4 plowings -137 75 -1.8 0.07

Rain distribution dummy (farmer assessment): 346 92 3.7 0.00
1 = good/excellent distribution

Gray soil dummy:1= gray color soil -216 72 -2.9 0.00
Source: Calculated from GMRP/MSU/SG2000/MOA Survey data.
Note: Adjusted R2 = 0.24.










Contribution of Environmental Factors on Teff Yields: A number of variables summarizing
information on environmental factors were used in the teff regression models. These variables
included soil type (distinguished by color, local names and clay/sand content), farmer assessment
of soil fertility, and farmer assessment of various biotic and abiotic stresses (e.g., rain shortage,
rain distribution, weed damage, diseases, insects). Only two of these environmental variables
proved to be significant: farmer assessment of rain distribution and soil color.

Yields on plots where farmers classified the distribution of rainfall as good or excellent were 350
kg/ha higher than plots where farmers said rain distribution was poor. Teff planted on gray soils
yielded 200 kg/ha less on average compared to teffplanted on black or red soils. These results
were consistent across both models.

Contribution of Management Practices on Teff Yields: Number of plowings was the only
management variable that appeared to have a significant impact on teff yield. However, unlike
maize, the number ofplowings had a negative effect on yields. On average, farmers who plowed
more than four times got about 137 kg/ha less than those who plowed less. Fortunately, only 41
farmers in the sample plowed more than four times. These results are also consistent across
Models 1 and 2 (Tables 15 and 16).

The regression results presented in Tables 15 and 16 give a rough picture of factors that appear to
have had the most important influence on teffyields in the 1997 growing season in East Shoa.
However, it is important to note that these models explain only a portion (24%) of the variability
in the yield data.


3.3.3. Graduate Farmers' Decisions Concerning Choice of Teff Technology

Table 17 disaggregates the various components of the teff technology package and reports the
number of graduates continuing each given practice. The reasons given by graduates for not
continuing a recommended practice are listed in Table 18.











Table 17. Graduate Farmers' Response to Recommended Teff Practices, East Shoa
Number of Graduates 60

Technology Component Percentage of Graduates in the Zone Continuing a
Given Technology
Improved seed 40

Recommended seeding rate 45

Recommended rate of DAP 65

Recommended rate of Urea 25


Source: Calculated from GMRP/MSU/SG2000/MOA Survey data.


Table 18. Most Common Reasons Given by Graduates in East Shoa for Not Continuing A
Given Teff Technology Component

% of Farmers Not Using the
Common Reasons for Discontinuing a Given Recommended Practice for a
Technology Component Technology Component Given Reason
Improved seeds Couldn't get improved seed 61
Seed isn't supplied/local seed is better 39
Seeding rate Seed damage necessitated second planting 55
To control weeding 30
DAP rate High cost of fertilizer 52
Land is fertile 19
Urea rate Urea causes plant lodging 51
High cost of fertilizer 25
Source: Calculated from GMRP/MSU/SG2000/MOA Survey data.


Farmers in East Shoa appear to have little confidence in the seed type and seeding rate
recommended by the NEP. Only 40% of the graduates surveyed continued to use Qoladma or
Magna seed-a rational decision, as shown above, since those who used retained seed obtained
significantly higher yields than Qoladma/Magna users (Table 17). The main reason given for
switching from the recommended seed variety back to retained seed was the unavailability of










improved seeds (Table 18).9 Graduates also nearly doubled the recommended seeding rate from
35 to 69 kg/ha (Table 17). Farmers said the seed they received was often damaged and it was
necessary to increase the seeding rate in order to avoid having to plant a second time. The other
main reason given for a higher seeding rate was to increase the plant density in order to control
weed growth.

Almost two-thirds of the graduates continued to use DAP at the recommended rates, but only
one-quarter continued to use high levels of urea-again, as shown above, this appears to be a
very rational decision. Decreasing urea rates by 50% had no significant impact on yields in
Models 1 or 2. Farmers cited the high cost of fertilizer as the most important reason for
discontinuing DAP and urea in general. The main reason given for reducing the urea rate was the
problem of lodging associated with urea application.































9 This reason reported by graduates does not seem quite convincing given the fact that Qoladma and Magna seeds
were available and used by some graduates and on traditional plots.











4. FINANCIAL ANALYSIS


Section 3 presents yield results and analyzed the key factors contributing to yield differences
across plots where farmers used different levels of inputs. These results demonstrate that the use
of improved seed and fertilizer technology significantly increased maize and teff yields in the
surveyed zones. Farmers using improved technology also incur additional costs to obtain these
yield increases, however. These include the cost of the inputs themselves, interest charges if the
inputs are obtained on credit, and the cost of additional labor that may be required for fertilizer
application, weeding, and harvest. This section analyzes whether it is financially profitable for
farmers to use improved technology for maize and teff, i.e., do the gains from sale of these
commodities compensate farmers for the costs of production?


4.1. Data and Methods Used

The study used two measures, net income per hectare and net income per labor day, to evaluate
financial profitability under different price and technology scenarios. Summaries of key results
from the maize and teff financial budgets are presented in Tables 19 and 20. Financial results are
reported by program type and input use level. All financial analysis results are reported in
Ethiopian birr (birr).10

Using plot-level data from the survey and additional secondary data, net income was calculated
as: (a) gross revenue was calculated by multiplying the crop yield per hectare by the farmgate
price"; and (b) costs of production reported by survey farmers or program administrators were
then subtracted from the gross revenue to obtain net income per hectare. These costs included
the cost of inputs such as seed, DAP, urea, herbicide, insecticide, and fungicide; interest costs on
input loans if applicable12; cash or the cash value of in-kind payments to non-family laborers
working on the plot; the depreciated value of animals and tools used for animal traction, including
animal feed and health maintenance costs; and the depreciated value of hand tools used in crop
production and the cost of sacks used to transport the commodity to market.



'1The average exchange rate during the 1997 crop and marketing year was US $1.00=6.70 birr.

11Wholesale prices from main market towns in each zone (East Shoa, West Shoa, Jimma) were obtained from the
Ethiopian Grain Trading Enterprise (EGTE) and adjusted to farmgate prices using data from our survey.
12Current MOA/SG maize program participants paid no interest on seed, fertilizer, and pesticide inputs while
current teff NEP program participants and maize/teff graduate and traditional farmers who received inputs through
the NEP program paid 10% interest annually. Under Ethiopian law a non-profit organization is not allowed to
earn income, hence SG2000 does not charge interest to farmers who participate in the MOA/SG demonstration
program.










In financial analysis no monetary value is imputed to family labor, but net income per day of
family labor is calculated by dividing the net income per hectare by the number of (family) adult
equivalent days used during crop production and harvest. Net income per day of family labor can
be compared to area wage rates (which approximate the opportunity cost of labor) to assess the
relative attractiveness of the technology at different yield and price levels.

We calculated net income under several different price scenarios, assuming that farmers
harvesting in November 1997 sold their crop in (a) January 1998, (b) April-May 1998, and (c)
August 1998, to assess potential gains from storage. In each case crop yields were adjusted to
reflect storage losses13 and interest charges according to the length of the loan period. Gross
revenue was also adjusted to reflect the opportunity costs associated with selling at different times
of the year. During 1998, actual maize and teff prices rose throughout the season. Net income
per hectare was also calculated for hypothetical drops in output prices of (d) 25%, and (e) 50%
from their January 1998 values. An additional scenario (f) was calculated for maize in which it
was assumed that farmers selling in August 1998 were able to cut their storage losses in half
through the purchase and use of storage insecticide.

A review of the budgets presented in Tables 19 and 20 led to three key conclusions about maize
and teff profitability. These are presented in Sections 4.2 through 4.4 below.


4.2. The Use of Improved Technology for Maize and Teff Is Extremely Profitable, Even if
Output Prices Decline by 25% or 50%

For both teff and maize, net income per hectare and per labor day are high (though variable
between yield terciles) for farmers using improved technology. Both measures of return were
extremely robust under all price scenarios. Net income and returns per labor day increase with
yield, but variable costs are covered and returns to family and mutual labor exceed average daily
labor rates (3-6 birr per day) under all price scenarios, for almost every program and input level
category.

The sole exception was the case of teff farmers using program-supplied seed with recommended
quantities of DAP and urea. With a hypothetical drop in output price of 50% below January 1998
levels, net returns remained positive but net income/day dipped to 1.6 birr/day, below average
wage rates. In all other cases returns were positive and net income/day exceeded wage rates by a
large margin.




13Teffis highly resistant to pests and it was assumed that no storage losses occurred (Seyfu 1993). Maize storage
losses were assumed to be 2% per month, the average of various estimates from Abraham et al. 1993.











4.2.1. Jimma-Maize


Results by Program Type: In Jimma, net returns per hectare and per labor day from program
plots were double those from traditional plots. Gains from graduate plots were even higher since
graduates had significantly higher yields than program participants while costs were the same or
lower. Even for the traditional plots net income was positive and returns per labor day exceeded
the average wage rate.

At January output prices, net income per hectare ranged from 1029 birr/ha for traditional plots to
2042 birr/ha for MOA/SG plots and 2543 birr/ha for graduate plots. Returns per labor day were
one-third higher on program plots (15 birr/day) compared to traditional plots (11 birr/day) at
January output prices. Graduate plot returns per day (18 birr/day in January) were 60% higher
than traditional plots on average. In the most pessimistic case, assuming a price drop of 50%
from January levels, net income per hectare falls to 293 birr/ha for traditional plots, to 601 birr/ha
for MOA/SG program plots, and to 768 birr/ha for graduate farmers. Net returns per labor day
decline to 3.2 birr/day (traditional), 4.4 birr/day (MOA/SG), and 5.5 birr/day (graduate).

Results by Input Level: Net returns per hectare for farmers who used the complete package of
inputs (improved seed plus DAP and urea >= or < the recommended rate) were roughly double
the returns received by farmers who used only local seed with DAP in every price scenario. At
January prices, net returns per hectare ranged from 1054 birr/ha (local seed + DAP) to 2107
birr/ha (improved seed, DAP, urea >= recommended rate) to 2261 birr/ha (improved seed, DAP,
urea <= recommended rate). Farmers in all groups still recover all variable costs and make a
profit even if price levels drop 50% below the January levels. Net returns for the case of a 50%
price drop range from 510 birr/ha (local seed + DAP) to 997 birr/ha (improved seed + DAP +urea
>= recommended rates) to 1135 birr/ha (improved seed + DAP + urea < recommended rate).

Returns per labor day were highest for farmers who used improved seed plus DAP and urea
below the recommended rate. Most farmers in this group were program graduates and used labor
more efficiently than new program participants. Returns per labor day for this group were 71-
80% higher than the group using local seed plus DAP, and 50% higher than returns for the group
using inputs at or slightly above the recommended rate. Farmers using improved seed and
fertilizer at >= the recommended rate had returns 13-16% higher than the local seed plus DAP
group. For all groups, returns per day from maize production far exceeded the prevailing daily
wage rate. Returns to maize production ranged from 11.3 (DAP plus urea) to 19.7 (improved
seed plus DAP, urea < recommended rates) for January 1998 sales, compared to prevailing wage
rates of 3-6 birr/day. If price levels drop by 50%, returns per day drop but are still above the
wage rate: they are 5.5 birr/day (local seed + DAP), 6.2 birr/day (improved seed, fertilizer >=
recommended rate), and 9.9 birr/day (improved seed, fertilizer < recommended rate).










4.2.2. West Shoa-Maize


Results by Program Type: Returns in West Shoa were also high and robust in response to price
changes, but results did not vary significantly between the two program groups. At January 1998
prices, MOA/SG participants achieved returns of 2781 birr/ha and graduates got 2702 birr/ha. A
price drop of 50% reduces gains to 940 birr/ha for MOA/SG participants and 1110 birr/ha for
program graduates.

Unlike Jimma, program graduates in West Shoa appeared to use labor less efficiently than current
program participants, but this is probably due to the fact that less than half continued to use
improved inputs after leaving the MOA/SG program, as discussed in Section 3. Although net
returns per hectare were similar between the two groups, returns per labor day were up to one-
third higher for current program participants. In January 1998 net returns per day were 17.6
birr/ha for MOA/SG plots compared to 13.1 birr/day for graduates, declining to 5.9 and 5.4
birr/ha respectively when output prices drop by 50%.

Results by Input Level: Under all price scenarios, net returns per hectare and per labor day were
highest for the group using improved seed and fertilizer at less than the recommended rate (mostly
program graduates). If maize is sold at January 1998 prices, net returns per hectare ranged from
2316 birr/ha (local seed only), to 2759 birr/ha (improved seed plus fertilizer >= recommended
rate) to 3102 birr/ha (improved seed plus fertilizer < recommended rate). The package of
improved inputs outperformed local seed with no fertilizer in all but the worst price scenario (-
50% from January 1998 levels). In that case, gains from improved seed plus fertilizer >=
recommended rates were 874 birr/ha compared to 1037 birr/ha (local seed, no fertilizer) and 1185
(improved seed, fertilizer < recommended rates).

Returns per labor day were highest for farmers using improved seed and fertilizer below the
recommended rates. Returns per day for this group were 23-47% higher than for the group using
inputs at >= recommended rates, and 50-74% higher than returns/day for farmers using only local
seed. At January 1998 price levels returns per day ranged from 11.4 birr/day (local seed only) to
16.0 birr/day (improved seed + fertilizer >= recommended rates) and 19.6 birr/day for the groups
using improved seed plus fertilizer at less than the recommended rates.


4.2.3. East Shoa-Teff

Results by Program Type: As discussed in Section 3, differences between program types in East
Shoa are blurred because teff farmers used improved varieties and fertilizer on all teff plots
(program, traditional, graduate). Yield differences between program types were not significant,
but net income per hectare was significantly higher for traditional and graduate plots than for
program plots. At January 1998 price levels, net returns ranged from 1904 birr/ha (MOA/SG) to
2091 birr/ha (traditional) and 2193 birr/ha (graduate). Net returns remained positive even










assuming a drop of 50% in January 1998 prices, with net returns declining to 366 birr/ha
(MOA/SG), 587 birr/ha (traditional), and 595 birr/ha (graduate).

Labor requirements for teff are much lower than for maize. Thus while net returns per hectare are
comparable or slightly lower than returns received by maize farmers in West Shoa and Jimma, net
returns per labor day were much higher because teff required only half as much (or less) family
labor. Net returns/labor day at January prices were 29.7 birr (MOA/SG), 36 birr (traditional), and
28.5 birr (graduate). If the teff price declines by 50% these returns drop to 5.7 birr (MOA/SG),
10.1 birr (traditional), and 7.7 birr (graduate)-still above average area daily wage rates of 3-6
birr/day.

Results by Input Level: These results are much more revealing. Because of an apparent problem
with the improved varieties of teff distributed through the MOA/SG program during the 1997
season, farmers using program seed with recommended quantities of fertilizer had significantly
lower yields than farmers using saved (improved) seed and fertilizer. Users of program-
distributed seed obtained net returns that were only two-thirds as high as returns in the other two
categories. At January 1998 prices, net returns ranged from 1331 birr/ha (program seed,
recommended quantities of fertilizer) to 2192 (saved seed, recommended quantities of fertilizer)
to 2306 (saved seed, recommended quantities of DAP, 50% of the recommended quantity of
urea). Farmers who used only half the recommended quantity of urea achieved almost the same
yields as farmers using more urea, but had higher net returns since their costs were lower. This
suggests that there is a need to review and possibly revise the fertilizer recommendation for teff in
this area to help farmers maximize income gains from the use of improved technology.

Net returns per labor day were also highest for farmers using saved (improved seed) with
recommended quantities of DAP and 50% of the recommended quantity of urea. Returns at
January 1998 prices ranged from 19.6 birr/day (program seed, recommended quantities of
fertilizer) to 32.7 birr/day (saved seed, recommended quantities of DAP, urea) and 34.9 birr/day
(saved seed, near recommended DAP, 50% urea). In the worst-case scenario, in which January
1998 prices drop by 50%, net returns per labor day drop to 1.6 birr/day for the program seed
group, below average wage rates for the area, but remain above the wage rate for the other two
groups: 7.8 birr/day (saved seed, recommended quantities of fertilizer) and 10.3 birr/day (saved
seed, recommended DAP, 50% urea).

43. Gains from Storage and Use of Storage Insecticide

4.3.1. Teff

There were significant gains from storing teff for later sale in the 1998 season. Farmgate grain
prices rose by 23% and straw prices doubled between January and August 1998. Farmers can
increase net income by more than 33-44% by selling in August instead of January. It should be
noted that current MOA/NEP contracts require farmers to pay back their input loans soon after











harvest, however. Since liquidity is a constraint for most farmers, unless the contracting
arrangements can be changed to permit farmers to repay the loans at a later time (with additional
interest charges) they may be unable to benefit from these income gains through storage.


4.3.2. Maize

Maize prices also rose markedly between January and August 1998 in West Shoa (29%) and
Jimma Zones (72%). Unlike teff, untreated maize deteriorates rapidly in storage. In Jimma, the
price rise over time was steep. Even accounting for storage losses, net income per hectare and
per labor day rose by over 60% between January and August. In West Shoa the price rise was
less dramatic, and net income increased by just 7-8% if farmers stored and sold maize in August
instead of January.

None of the survey farmers reported using storage insecticide following the 1997 production year,
but our results indicate that income gains from pesticide use (through reducing the amount of
maize lost to storage pests) would be substantial in both Jimma and West Shoa. If Jimma and
West Shoa farmers used insecticide and storage losses were reduced by half14, net income per
hectare would increase by 19-23% if farmers sold in August 1998 rather than January, even after
the costs of storage insecticide are deducted. Net income for Jimma farmers (selling in August
rather than January) would increase by 80-85%.


4.4. Improved Seed and Fertilizer Costs Represent 50-75% of Total Costs

Improved seed and fertilizer are by far the biggest cost component in the financial enterprise
budgets. In East Shoa (teff) the costs of improved seed and fertilizer represent more than half of
total costs (exclusive of family labor). Purchased seed and fertilizer make up two-thirds to three-
quarters of total production costs for maize in West Shoa and Jimma. This suggests that even
small reductions in the farmgate cost of fertilizer and seed (e.g., by reducing transport and other
marketing costs) could significantly increase farm profits.










14This is a conservative estimate. Recent research suggests that the application of storage insecticide can reduce
storage losses to 2-13% of grain weight over a 5-9 month period (Abraham et al. 1993).
















Table 19. Summary of Maize Results: Financial Analyses by Zone, Program Type and Input Level
JIMMA WEST SHOA
Program Type Input Level Program Type Input Leve
Imp. Seed Imp. Seed
+ DAP+ Imp. Seed + DAP+ Imp. Seed
Urea + DAP+ i Local Urea + DAP+
MOA/ Tradi- Local Seed = MOA/ Tradi- Seed, No =
Zone/Budget Item SG tional Graduate + DAP Rate Rec. Rate SG tional Graduate i Fertilizer Rate Rec. Rate
YIELD (t/ha) 1/ 5.6 2.8 6.8 2.9 6.0 5.9 5.6 n/a 4.8 3.9 5.8 5.7
TOTAL FAMILY/MUTUAL LABOR DAYS
(adult equiv. days/ha) 135 92 140 93 115 162 159 206 204 158 172
N used in calculations 69 47 39 [ 43 58 50 92 57 33 45 68
FINANCIAL ANALYSIS
a. Net Income (Birr/ha) 2/
Jan 98 Price 2042.1 1029.1 2543.2 1053.7 2260.8 2106.7 2781.0 2702.4 2316.1 3102.3 2758.9
April-May 98 Price 2300.8 1160.3 2848.3 1201.0 2564.7 2404.7 2577.7 2521.0 2184.1 2902.9 2562.4
August 98 Price 3257.4 1648.0 4012.6 1715.1 3626.3 3450.5 3010.9 2890.1 2477.2 3340.7 2991.9
Aug 98 w/storage insecticide 3577.2 1811.0 4405.7 1937.2 4082.6 3901.3 3322.0 3159.2 2759.2 3761.4 3405.5
Jan 98 Price 25% 1321.3 660.8 1655.8 990.9 2130.9 1978.7 1860.3 1906.2 1676.5 2143.4 1816.5
Jan 98 Price 50% 600.5 292.5 768.4 509.5 1135.2 997.2 939.6 1110.0 1037.0 1184.6 874.1
b. Net Income per Family and Mutual Labor Day (Birr/ha) 3/
Jan 98 Price 15.1 11.2 18.2 11.3 19.7 13.0 17.6 13.1 11.4 19.6 16.0
April-May 98 Price 17.0 12.6 20.3 12.9 22.3 14.8 16.3 12.2 10.7 18.4 14.9
August 98 Price 24.1 17.9 28.7 18.4 31.5 21.3 19.1 14.0 12.1 21.1 17.4
Aug 98 w/storage insecticide 26.5 19.7 31.5 20.8 35.5 24.1 21.0 15.3 13.5 23.8 19.8
Jan 98 Price 25% 9.8 7.2 11.8 10.7 18.5 12.2 11.8 9.3 8.2 13.6 10.6
Jan 98 Price 50% 4.4 3.2 5.5 1 5.5 9.9 6.2 5.9 5.4 1 5.1 7.5 5.1
Sources: Survey and secondary data.
1/Estimated from crop cuts. Maize assumes storage losses of 2%per month. Yield differences were significant at the 95% level between all groups within the program and input level categories EXCEPT
improved seed, DAP, urea < recommended rate and improved seed, DAP, urea >= recommended rate in both Jimma and West Shoa.
Teff assumes no grain or straw lost during threshing and no storage loss. Yield differences between program groups were not significant. For input level groups, yield differences between program seed and
saved seed categories were significant at the 95% level. Differences between the saved seed groups were not significant.
2/All prices are in Ethiopian birr. During the study period the average exchange rate was USD 1 = 6.7 birr. Net Income=Gross revenue (cash costs + interest + purchased labor + animal traction costs + hand
tool and sack cost). Prices used were as follows (birr/kg) (from EGTE and survey data):
Maize -Jimma: Jan. .54, April-May .65, Aug. .93; Maize-West Shoa: Jan. .69, Apri-May .72, Aug. .89
Teff -East Shoa: Jan. 2.04, April-May 2.11, Aug. 2.51
3/Net income/total family and mutual labor days















Table 20. Summary of Teff Results: Financial Analyses by Zone, Program Type and Input Level
EAST SHOA
Program Type Input Level
S Prog. Seed, Saved (Imp.) Seed, Near
SRecommended Quantities Saved (Imp.) Seed, Near Recommended DAP, 50%
Zone/Budget Item MOA/SG Traditional Graduate DAP, Urea Recommended DAP, Urea Urea
GRAIN YIELD (t/ha) 1/ 1.4 1.4 1.5 1.1 1.5 1.5
STRAW YIELD (t/ha) 1/ 2.2 2.0 2.1 2.1 2.1 2.1
TOTAL FAMILY/MUTUAL LABOR DAYS 64.0 58.0 77.0 68.0 67.0 66.0
N used in calculations 60.0 60.0 60.0 _35.0 63.0 69.0
FINANCIAL ANALYSIS
a. Net Income (Birr/ha) 2/
Jan 98 Price 1903.6 2090.5 2193.4 1331.4 2192.1 2306.0
April-May 98 Price 2008.9 2192.6 2299.5 1431.6 2385.0 2494.0
August 98 Price 2602.7 2771.9 2912.5 1912.0 3139.7 3227.8
Jan 98 Price 25% 1134.6 1338.6 1394.0 721.8 1356.4 1493.8
Jan 98 Price 50% 365.6 586.8 594.5 112.1 520.7 681.6
b. Net Income per Family and Mutual Labor Day (Birr/ha) 3/
Jan 98 Price 29.7 36.0 28.5 19.6 32.7 34.9
April-May 98 Price 31.4 37.8 29.9 21.1 35.6 37.8
August 98 Price 40.7 47.8 37.8 28.1 46.9 48.9
Jan 98 Price 25% 17.7 23.1 18.1 10.6 20.2 22.6
Jan 98 Price 50% 5.7 10.1 7.7 1.6 7.8 10.3
Sources: Survey and secondary data,
1/Estimated from crop cuts. Maize assumes storage losses of 2%per month. Yield differences were significant at the 95% level between all groups within the program and input level categories EXCEPT
improved seed, DAP, urea < recommended rate and improved seed, DAP, urea >= recommended rate in both Jimma and West Shoa.
Teff assumes no grain or straw lost during threshing and no storage loss. Yield differences between program groups were not significant. For input level groups, yield differences between program seed and
saved seed categories were significant at the 95% level. Differences between the saved seed groups were not significant.
2/All prices are in Ethiopian birr. During the study period the average exchange rate was USD 1 = 6.7 birr. Net Income=Gross revenue (cash costs + interest + purchased labor + animal traction costs + hand
tool and sack cost). Prices used were as follows (birr/kg) (from EGTE and survey data):
Maize -Jimma: Jan..54, April-May .65, Aug. .93; Maize-West Shea: Jan. .69, April-May .72, Aug. .89
Teff-East Shea: Jan. 2.04, April-May 2.11, Aug. 2.51
3/Net income/total family and mutual labor days











5. ECONOMIC ANALYSIS


In the preceding section we considered whether or not it is financially profitable for farmers to
use improved maize and teff technology, given the farmgate prices they pay for inputs, other
production costs, and the farmgate prices received for maize and teff output. This section
summarizes the results of the economic analysis, which considers profitability from the viewpoint
of society instead of the individual farmer. Three questions are addressed:

0 (1) Which is cheaper for society overall-producing maize and teff
domestically, using improved technologies, or importing foodgrains?

D In the past Ethiopia has mainly been a cereal grains importer, but in two of the past
five years it has exported grain to neighboring countries. If production gains
continue, it may become common for Ethiopia to have grain surpluses that could
be exported. (2) Will it be socially profitable for Ethiopia to export maize and
teff produced with improved technology, i.e., will the economic returns from
exports cover the economic costs?

D The Ethiopian government is currently subsidizing the costs of extension assistance
and credit provision for program farmers, but the price farmers pay for improved
seed and fertilizer does not include the full cost of these programs.
(3) Do the benefits of government-supported programs to facilitate
technology adoption outweigh their costs to society?

The following sections review the differences between financial and economic analysis and explain
the procedures followed for calculating economic values in the Ethiopia case. The results are
presented and discussed, and are summarized in Tables 22 through 24. A number of scenarios are
considered: (1) net benefits when maize and teff produced through the intensive production
program are valued as import substitutes, and world fertilizer prices are high; (2) net benefits
when maize and teff are valued as import substitutes, and world fertilizer prices are low; (3) net
benefits when maize production through the program is assumed to be exported at a relatively
high price, and farmers face high world fertilizer prices; (4) break-even maize export prices when
fertilizer prices are high; (5) net benefits when maize and teff are valued as either import
substitutes or exports and costs of operating the extension and credit programs are added; and (6)
net benefits when extension and credit program costs are reduced by half.


5.1. Differences Between Financial and Economic Analysis

The economic analysis considers the impact of technology program costs and benefits on society
as a whole. In practice, this involves adjusting financial prices to economic prices in two steps.










First, taxes and subsidies, which were included in the financial analysis, are excluded in economic
analysis because they represent transfers of funds within the economy from one group to another
group, via the government, without affecting total national income."1 For example, the Ethiopian
government used to provide a direct subsidy on the price of fertilizer to farmers. The price subsidy
has now been removed, but if it were still in place, this type of subsidy would be considered a
direct income transfer (here, from the government or taxpayers to producers) rather than a real cost
reduction, and would not be included in the economic analysis.

The additional costs incurred by the extension agency to provide intensive technical assistance to
program participants and manage the MOA/SG credit program, or to provide other support
services, should be treated as a cost to the society, even if farmers do not pay for them directly,
since these expenditures reflect the use of real resources to support the technology program.

The second step in the economic analysis is to calculate the appropriate values for traded items
(e.g., imported inputs, import substitutes, and exports). Traded items are expressed in terms of
their world price equivalents. These are often termed "border prices," since they are based on
import or export prices, but they are often calculated at the farm level by adjusting for domestic
transportation and marketing costs between the point of import or export and the production
zone. In addition, the values of traded items are converted from foreign currency into domestic
currency using the economic foreign exchange rate (in our Ethiopia study, the parallel market
exchange rate) rather than the official exchange rate (in Ethiopia, the domestic auction rate). The
resulting prices are referred to as import or export "parity" prices.


5.2. Method Used to Determine the Economic Values of Traded Items

Four steps were followed: (1) estimating the parallel exchange rate; (2) establishing what
proportion of costs represent tradeable items; (3) converting that amount to local currency terms
using the parallel exchange rate; and (4) estimating the import parity prices for maize, teff, and
fertilizer and substituting the import parity prices for market prices in the economic analysis.
Ethiopia has been a net cereal importer in most years, but exported maize to Kenya in two of the
past five years. Therefore, an export parity price is also estimated for maize.


5.2.1. Parallel Exchange Rate

Parallel exchange rates are reported by the National Bank of Ethiopia. On average the birr was
overvalued by 3-10% for the period of our analysis. In calculating the import parity price of teff


'5The process of imposing taxes and subsidies distorts markets and therefore entails economic costs. Where it is
possible to estimate the magnitude of these losses, they should be taken into account.










and maize, we used the average rate for October 1997-August 1998, the period when imported
grains (substituting for domestic production in the crop year 1996/97) would be ordered and
delivered to Ethiopia. Seed and fertilizer inputs for the 1996/97 crop season would have been
ordered and imported earlier: the average parallel rate for November 1996-March 1997 was used
for the input import parity calculations.


5.2.2. Import and Export Parity Prices

Import parity prices for grains (calculated at the farm-level) estimate the maximum price which, if
paid to farmers in the different zones, would be comparable to the full cost of grain imported from
the United States. Since teff is not widely traded on the world market, the import parity price for
wheat, a substitute for teff in Ethiopia, is calculated instead. Because there is a significant price
difference between teff and wheat in the domestic market, however, a price premium of 40%
(reflecting the higher value consumers place on teff over wheat) was added to the wheat price
based on price data from the FEWS-European Union Food Security Project.

Details of the calculation of the import parity prices for maize, teff, and fertilizer are available at
http://aec.msu.edu/agecon/fs2/papers/index.htm#recentidwp. It is assumed that maize and wheat
imported to Ethiopia would come from the U.S. and the FOB price at the point of export from the
U.S. Gulf, for shipment to either Assab or the port of Djibouti, is used as the basis for the
calculation.

It is also assumed that DAP fertilizer would be imported from the U.S., but urea would come
from Middle Eastern sources. Because world fertilizer prices can vary substantially from year to
year (especially nitrogen-based fertilizers), economic prices are calculated for two scenarios:
"low" and "high" world fertilizer prices.16 Transportation to the port and insurance costs are
added to get the CIF price at the port of entry. Estimates of shipping, port, transportation and
handling costs were provided by a private Ethiopian fertilizer importer (Kassahun 1998). The CIF
price is adjusted by the cost of transport and marketing to the farmgate to get a farm-level
economic import parity price for each commodity. Transport and marketing costs were estimated
using data collected by our survey supervisors and secondary data. The financial transportation
rates are adjusted to economic prices by assuming that 75% of the cost of rail and truck transport
is composed of tradeable goods and valued at the parallel exchange rate.
Because maize seed is a tradeable item, the economic price of maize seed was based on the price
charged by Pioneer Hi-Bred International Seed Company. Pioneer imports basic seed for hybrid
maize from Zimbabwe and multiplies it in Ethiopia. The price for hybrid maize seed charged by


16The "low" DAP price was $200/ton (FOB U.S. port); the high DAP price was $240/ton. Low and high urea
prices were $100/ton (FOB Mideast port) and $225/ton. The price range was based on an analysis of fertilizer
originating from several U.S., European and Middle Eastern ports in 1997 and 1998 (Stepanek 1999).










Pioneer-Ethiopia seems to cover the full costs of seed production and marketing, unlike the price
charged by the Ethiopian Seed Enterprise (ESE), which supplied the MOA/SG program. ESE's
production costs are subsidized by the government of Ethiopia.

In calculating the export parity price for maize, we assume that the most likely market for future
exports is Kenya. The export parity price shows, for a given CIF price offered in Mombasa, what
price could be paid to Ethiopian farmers in different zones after shipping, road transport and
marketing costs are subtracted.


5.2.3. Estimating Program Costs

The study assumes that program costs for the MOA/SG and NEP programs are similar (personal
communication, Quinones, September 1998). Estimated costs of extension assistance and credit
administration are based on the 1995 NEP program budget reproduced below in Table 21 (MOA,
cited in Gordon, Habtemariam, and Kiflu 1995). The budget provides costs only for the NEP
program, not total costs for the extension agency. For example, all staff salaries (NEP and non-
NEP) and operating costs for non-NEP programs are excluded. The costs of credit administration
are also included in the budget, since credit provided to participating farmers was administered by
the extension service through the 1996/97 season. Extension will continue to manage the credit
program for the next year or two, after which banks and cooperative societies are expected to
take it over (Gordon, Habtemariam, and Kiflu 1995).

The study therefore assumes that the budget is a fair approximation of the incremental cost of the
MOA/SG and NEP programs, i.e., that the items represent extra costs incurred by the new
program, not basic costs of the day-to-day running of the extension agency that would have been
supported by the government anyway.











Table 21. NEP Extension Intervention Budget for the 1995 Crop Season
Item Value in Ethiopian Birr Share

Input cost 11,370,123 24%

Mobility (Lorry, 18,364,242 39%
Vehicle, Motors,
Bicycle)

Training and field 1,123,225 2%
equipment

Operating costs 9,916,323 22%

Tariffs and bank 6,269,626 13%
commission

Total 47,043,840 100%
Source: MOA, reproduced in Gordon, Habtemariam, and Kiflu 1995.


The study calculated a per-demonstration plot/farm estimate of program costs as follows:

S There were 35,160 demonstration plots planned for 1995 (MOA, cited in Gordon,
Habtemariam, and Kiflu 1995).

* The estimated total cost per demonstration plot: 47,043,840/35160 = 1338 birr per
demonstration plot.

* We subtracted input, bank charges, and tariffs since these are already accounted for or should
be excluded from the economic budget = 47,043,840 11,370,123 6,269,626 = 29,404,091
birr. 29,404,091/35160 = 836 birr per demonstration plot or farm.

* The budget is 39% transport-related, 2% training and field equipment. We assumed that 75%
of these category totals were traded goods and valued them at the parallel exchange rate.
Revised cost per farm = 843 birr.


5.3. Economic Analysis: Summary of Main Findings

Key results from the economic analysis are summarized in Tables 22 through 24. These tables
summarize results by zone, program type and input level. More detailed economic budgets for










maize, teff, fertilizer and seed, including notes on specific data sources can be found at
http://aec.msu.edu/agecon/fs2/papers/index.htm#recentidwp. The key findings of the economic
analysis are discussed below in Sections 5.3.1-5.3.3.


5.3.1. Economic Profitability of Intensive Maize and Teff Production

If domestically produced maize and teff will substitute for commercial imports, it is highly
profitable from society's standpoint to produce these crops using intensive technology. The net
economic benefits of domestic production of improved maize and teff are high and stable even
when world fertilizer prices are high.

Maize Import Parity with High/Low Fertilizer Prices: Assuming that increased domestic
production substitutes for imports, net gains to society from the use of improved inputs on maize
in Jimma exceed gains from more traditional practices by 100-114% (Table 22, lines a-b). Net
gains when farmers use local seed plus DAP were 3067-3102 birr/ha, depending on the
assumption of low or high world fertilizer prices. Farmers using improved seed and recommended
or higher-than-recommended fertilizer rates earned 6124-6289 birr/ha. Those using improved
seed and fertilizer at less than the recommended rate (mainly program graduates) had the highest
net gains, 6515-6638 birr/ha.

For West Shoa, the gains to society from the use of improved inputs instead of more traditional
practices were less dramatic but still sizable. Again, the payoff was highest for experienced
farmers who used less-than-recommended fertilizer rates. This group had net gains 43-46% higher
(6731-6853 birr/ha) than farmers using only local seed with no fertilizer (4691 birr/ha). Net gains
for program participants using fertilizer at recommended rates or higher were 32-36% (6198-
6370 birr/ha) greater than the low-input group.

The net nominal protection coefficient (NPC) for maize, defined as the financial price divided by
the economic price," ranges from .34 to.35 in Jimma and from .45 to .49 in West Shoa. A NPC
less than one indicates that the market price actually received by farmers for maize is less than its
economic value and therefore acts as an implicit disincentive to farmers.

TeffImport Parity with High/Low Fertilizer Prices: As with maize, net gains to society from the
production of teff with improved inputs are high and robust even when fertilizer prices are
relatively high, if teff is valued as an import substitute (Table 23, lines a-b). This analysis also
shows the impact of the apparent problem with teff seed distributed with the MOA/SG program in
1998. Farmers using improved seed provided by the program with recommended quantities of


17Here we compare the January 1998 farmgate financial price (Table 19) with the economic import parity farmgate
price assuming high fertilizer prices (Table 22).











DAP and urea had positive net gains, but at levels 75-102% (943-1084 birr/ha) lower than other
technology packages. Farmers who used saved (also improved) seed, near recommended levels of
DAP, and 50% of the recommended urea rate had the highest net gains (1908-1990 birr/ha), 5-
9% higher than individuals using saved seed and recommended rates of DAP and urea (1757-
1894 birr/ha).

Unlike the case of maize, the net NPC for teff8 exceeds 1, ranging from 1.21-1.41. This suggests
that teffs domestic price is quite high relative to its estimated value to society as an import
substitute. The magnitude of net gains to society from investments in improved teff technology is
also much lower (943-1990 birr/ha) than the societal payoff to investments in improved maize
technology (5783-6853 birr/ha) if cereal production is valued as an import substitute.


5.3.2. Variable Profitability of Maize for Export

Whether intensified maize produced for export is profitable at the societal level depends on the
prevailing export price. Our analysis looks quite different if we assume that maize produced with
purchased inputs will be exported rather than serving as an import substitute. This idea is not far-
fetched considering that maize production in Ethiopia has increased an average of 14% annually
from 1993 to 1998 (although annual levels fluctuated considerably) (FAOSTAT 1999). Imports
are declining and made up less than 2% of the total maize supply between 1993 and 1996
(FAOSTAT 1999). These data suggest that Ethiopia is at or nearing self-sufficiency levels in
maize.

Maize production will continue to rise as the NEP program expands. Average harvested maize
area in the period 1996-1998 was 1,683,000 hectares, and average yield was 1.7 tons/ha.
Assuming that the NEP program expands successfully in the future and farmers obtain 5 ton/ha
yields (MOA/SG averages) on 25% of total maize area (and 1.7 tons/ha on the remaining area),
maize production will increase to 4.2 million tons from current levels of 2.8 million tons (1996-
1998). If farmers achieve slightly lower yields of 4 tons/ha on 50% of maize area, total maize
production will rise to 4.8 million tons.

A similar projection can be made for teff. If teff yields increase to MOA/SG levels of 1.5 tons/ha
on 25% of total teff area, and farmers get 0.9 ton/ha (the national average yield) on the remaining
area, total teff production will rise from current levels of 2 million tons (1996/97) to 2.3 million
tons. If farmers can get slightly lower yields of 1.4 tons/ha on 50% of total teff area, production
will rise to 2.5 million tons.



1XComparing (as above) January 1998 fanngate financial price (Table 19) with the economic import parity
farmgate price assuming high fertilizer prices (Table 23).











High Export Price for Maize: If maize is produced primarily for export, the farmgate price is the
CIF price at the border of the importing country minus the costs of transport, marketing and
handling incurred in bringing the maize from the farmgate to the importing country's port. Under
these conditions, whether intensive maize is profitable or not from society's point of view will
depend on the export price. In 1997, Ethiopia exported maize to Kenya at a price of $194/ton CIF
Mombasa (Jayne, personal communication). At this price level, Table 22 shows that low-input
and intensive maize production remains profitable for maize farmers in Jimma and West Shoa,
although net gains are much lower compared to when maize is valued at import parity levels.

Net gains (at the $194/ton price'9) in Jimma range from 745 birr/ha for traditional farmers to 1972
birr/ha for program graduates. Net gains were 87-130% higher in groups using improved seed
with fertilizer (1429-1752 birr/ha) compared to farmers using local seed plus DAP (764 birr/ha).

Because farmers in West Shoa obtained relatively high yields even without improved seed and
fertilizer (3.9 tons/ha, compared to 5.7-5.8 tons/ha for improved seed and fertilizer users), the
profit gain from using improved inputs on maize (at $194/ton CIF Mombasa) is less striking. Net
gains from the use of local seed with no fertilizer (1635 birr/ha) are nearly the same as net gains
for the group using improved seed and fertilizer at or exceeding the recommended rate (1695
birr/ha). Farmers who reduced the fertilizer rate fared better, with net gains of 2149 birr/ha.

Break-even Export Price: Considered to be an unusually high price, $194/ton CIF Mombasa is
influenced by a maize crop failure that year in Kenya. Future export prices are likely to be lower.
For example, the 1996-1998 average price FOB U.S. Gulf was $124/ton, and prices have trended
lower in the last two years. Shipping and insurance costs to East Africa are about $37/ton, so the
U.S. could supply maize CIF Mombasa at $161/ton or less.

Table 24 (line a) shows the export prices (CIF Mombasa) that maize farmers would need in order
for the investment in technology to break even from society's standpoint (net gain=0). In West
Shoa, break-even export prices range from $133/ton (CIF Mombasa) for farmers using local seed
and no fertilizer to $151/ton for farmers using improved seed and fertilizer at or above
recommended levels. Break-even export prices for Jimma are even higher: Jimma is farther away
from the ports of Assab/Djibouti and transport costs are higher. Break-even prices range from
$153/ton for farmers using improved seed and fertilizer at less than recommended rates to
$160/ton for farmers using improved seed and fertilizer at higher raises. This finding raises
questions about whether Ethiopian maize will be able to compete with lower-cost producers in the
regional market

Role of Transport Costs: Subsector marketing and transport costs for maize and inputs will have
to be reduced for Ethiopian maize to compete effectively on regional export markets, and in order


19Assumes high world fertilizer prices.











to provide Ethiopian farmers with a farmgate price that makes it attractive to produce surplus
maize. Transport costs between the port and farmgate add 22-57% to the CIF cost of DAP and
urea in West Shoa, 27-68% in Jimma, and 22-58% in Debre Zeit.


5.3.3. Extension and Credit Costs Reduce Profitability

Accounting for the cost of extension and credit reduces the economic profitability of teff
significantly; the impact on maize depends on whether it is an import substitute or an export. In
the economic analysis, net income represents the residual return to factors that facilitate crop
production but are not explicitly costed out in the analysis, such as (in this case) the costs of
implementing extension and credit programs. As a result of fertilizer market liberalization the
price of fertilizer is no longer directly subsidized by the government. This is confirmed by our
finding that the import parity price of fertilizer at the farmgate level (accounting for transport
costs) is close to the price farmers currently pay for DAP and urea through the SG and NEP
programs.

However, the government and donors are still facilitating farmer access to fertilizer in other ways,
i.e., through the provision of intensive extension assistance and credit programs. These program
costs are not recovered through the prices farmers currently pay for fertilizer: an analysis of
program costs in Section 5.2.3 suggests that the handling margin and interest on fertilizer sold
through the SG and NEP programs is insufficient to cover the costs of credit provision. The
study estimates the per farm cost of extension assistance and credit program administration to be
843 birr per program farmer. Accounting for these costs in the analysis reduces the economic
profitability (i.e., the net gains to society) of improved technology use, but the magnitude of the
impact differs by crop and whether the cereal produced is an import substitute or an export. Table
22 (lines c, d, f), Table 23 (lines b, c) and Table 24 (lines b, c) show the impact when program
costs are deducted from net gains under different fertilizer price and technology assumptions.

Estimates of program costs are based on MOA budgets from 1995. Between 1995 and 1997 (the
year of the survey), the number of farmers covered in the program expanded considerably. Since it
is unlikely that the program budget also increased significantly, costs per farmer probably
declined. Tables 22-24 show the impact of program costs on economic profitability if they are
only one-half of our original estimate, 422 birr/farmer.

Maize: If maize produced is considered to be an import substitute, the relative costs of extension
and credit provision are low and have little impact on the net gains to society. When maize is an
import substitute, accounting for program costs reduces net gains by only 11 to 15% in Jimma
and West Shoa. If program costs are reduced by 50%, net gains are reduced by only 5-8%.

The results are very different when the study assumes that intensive maize will be exported.
Under the export scenario net gains are much lower because the farmgate price for exported











maize is relatively low (even assuming that the CIF Mombasa price is $194/ton). Deducting
program costs reduces net gains of exported maize by 39-60%. If program costs are halved, net
gains are still reduced substantially, by 20-30% (Table 22).

Teff: As in exported maize, accounting for the costs of extension and credit severely reduces net
gains from intensive teff production-by 42-89%----depending on assumptions about fertilizer
prices and levels of inputs used. The impact is still substantial, a 21-45% reduction of net gains, if
program costs are reduced by 50%.

The results presented in Section 4 confirm the importance of intensive extension assistance and
on-time, reliable delivery of inputs in raising yields and motivating farmers' continued use of
improved technology. Government and donor support of these special programs is particularly
important at this time because there are few alternative channels for input supply and extension
advice. However, in the past in Ethiopia and in other countries, promising increases in input
utilization have ended abruptly with the withdrawal of special programs such as MOA/SG and
NEP. Identifying strategies to reduce program costs while maintaining the high quality of
extension and credit provision will be critically important if Ethiopia is to sustain this success
story among better-off farmers in Ethiopia and expand program coverage to poorer farmers. The
remainder of this paper discusses problems and key challenges for Ethiopia in reducing these
program costs and expanding the cereals intensification program.















Table 22. Summary of Economic Analysis Results for Maize by Zone, Program Type and Input Level
JIMMA WEST SHOA
Program Type Input Level Program Type Input Level
Imp.Seed
+ DAP+ Imp. seed + Imp. Seed + Imp.Seed +
Urea DAP+ Local Seed, DAP+ DAP+
MOA/ Tradi- Local Seed < Rec. Urea >= MOA/ No Urea Urea >= Rec.
Zone/Budget Item SG tional Graduate + DAP Rate Rec. Rate SG Graduate Fertilizer ECONOMIC ANALYSIS

Net Income (Birr/ha)

a. Import Parity Hi Fert Price 1/ 5783 2966 7325 3067 6515 6124 6182 5571 4691 6731 6198

b. Import Parity Lo Fert Price 2/ 5929 2998 7462i 3102 6638 6289 6334 5628 4691 6853 6370

c. Import Parity Hi Fert Price incl. extension,
credit costs 3/ 4940 n/a 64821 n/a 5672 5281 5339 4728 n/a 5888 5355

d. Import Parity Lo Fert Price incl. extension,
credit costs 3/ 5086 n/a 6619i n/a 5795 5446 5491 4785 n/a 6010 5527

e. Import Parity Lo Fert, 50% ext., credit costs 4/ 5361 69031 6093 5702 5760 5150 6310 5777

f. Import Parity Hi Fert, 50% ext., credit costs 4/ 5507 7040i 6217 5867 5913 5207 6432 5948
------------------- ---------------------------------------
g. Export Parity Hi Grain and Hi Fert. Prices 5/ 1435 745 19721 764 1752 1429 1763 1750 1635 2149 1695

h. Export Parity Hi Grain and Hi Fert. Prices incl.
extension, credit costs 592 n/a 1129 n/a 909 586 920 907 n/a 1306 852


Sources: Survey and secondary data.
1/Assumes the following fertilizer prices: DAP (FOB U.S. Gulf) USD 240; urea (FOB Middle East port) USD 225.
2/Assumes the following fertilizer prices: DAP (FOB U.S. Gulf) USD 200, urea (FOB Middle East port) USD 100.
3/Net income at the import or export parity price, assuming high fertilizer prices and accounting for costs of extension and credit program administration estimated at 843 birr per hectare.
4/Assumes extension and credit program costs are 422 birr/ha.
5/Assumes the maize price CIF Mombasa is USD 194 (T. Jayne, personal communication).














Table 23. Summary of Economic Analysis Results for Teff by Zone, Program Type, and Input Level
EAST SHOA
Program Type Inut Level
Prog. Seed, Saved (Imp.) Seed, Saved (Imp.) Seed,
Recommended Quantities Near Recommended Near Recommended
Zone/Budget Item MOA/SG Traditional Graduate DAP, Urea DAP, Urea DAP, 50% Urea
ECONOMIC ANALYSIS
Net Income (Birr/ha)

a. Import Parity Hi Fert Price 1/ 1498 1728 1750i 943 1757 1908

b. Import Parity Lo Fert Price 2/ 1640 1822 1846i 1084 1894 1990

c. Import Parity Hi Fert incl. extension, credit costs 3/ 655 n/a 907 100 914 1065

d. Import Parity Lo Fert incl. extension, credit costs 3/ 797 n/a 1003 241 1051 1147

e. Import Parity Hi Fert incl. 50% extension, credit costs 4/ 1077 13281 521 1335 1486

f. Import Parity Lo Fert incl. 50% extension, credit costs 4/ 1218 1425 662 1472 1569
Sources: Survey and secondary data.
1/Assumes the following fertilizer prices: DAP (FOB U.S. Gulf) USD 240; urea (FOB Middle East port) USD 225.
2/Assumes the following fertilizer prices: DAP (FOB U.S. Gulf) USD 200, urea (FOB Middle East port) USD 100.
3/Net income at the import or export parity price, assuming high fertilizer prices and accounting for costs of extension and credit program administration estimated at 843 binr per hectare (farm). Extension and
credit costs were calculated from Ministry of Agriculture Extension Department budgets reproduced in Gordon, Habtemariam, and Kiflu 1995.
4/Assumes extension and credit program costs are 422 birr/ha.















Table 24. Break-Even Export Prices for Maize
JIMMA WEST SHOA
Program Type Input Level Program Type Input Level
Imp. Seed Imp. Seed
+ DAP+ Imp. Seed + DAP+ Imp. Seed
Urea + DAP+I Local Urea + DAP+
MOA/ Tradi- Local Seed MOA/ Tradi- Seed, No < Rec. Urea >=
Item SG tional Graduate + DAP Rate Rec. Rat SG tional Graduate Fertilizer Rate Rec. Rate
a. Break-Even Export Parity Price CIF/Mombasa
assuming high fertilizer price ($/ton) 157 157 1531 157 153 160 147 1401 133 140 151

b. Break-Even Export Price, Hi Fertilizer
Price, including credit/extension costs ($/ton) 179 170 172 180 169 167 161 172

c. Break-Even Export Price, Hi Fertilizer
Price, including 50% of credit/extension costs ($/ton) 168 162 162 169 159 154 I 151 161
I 1










6. SUSTAINING THE MOMENTUM


The results presented in Sections 3 and 4 indicate that the use of improved technology
significantly increases yields and income for participant farmers. Results in Section 5 show that as
long as increased production can be consumed locally, MOA/SG programs to promote these
improved technologies are economically (i.e., socially) profitable after adjustments are made for
price distortions and government funding of extension and credit services. These results support
the GOE's current policy of encouraging sustained use of these new technologies by the early
adopters and extending their use to a broader base of farmers.

Nevertheless, the recent transition from the relatively small scale MOA/SG program (<3,500
participants at its height in 1995) to the much more ambitious goals of the NEP (almost 3,000,000
participants anticipated in 1998) reveals that there are numerous challenges to meet if this
momentum is to be sustained. Some of the more important challenges identified by field surveys
conducted in August 1998 are:

0 Expanding NEP extension coverage to a wider base of farmers without (1) diluting the
quality of the extension message, and (2) increasing program costs.
D Finding low-cost, financially sound ways of administering input credit.
D Developing a transparent and responsive input supply system that can function
independently of the NEP and provide all farmers access to low-cost, high-quality
agricultural inputs.


6.1. Expanding the NEP Extension Coverage to a Broader Group of Farmers

The objectives of NEP are similar to those of MOA/SG: to introduce farmers to new
technologies in a well-supervised environment so that after a year or two of participation in the
program they (1) see the merits of the new technology, and (2) are able to "graduate" and
continue using the new technology on their own without special assistance. The main difference
between the two programs is the scale. Analyses in Section 2, however, showed that MOA/SG
participants are wealthier than average farmers, with larger land holdings and better access to
resources such as animal traction and labor. This means that in addition to increases in the scale of
operations the NEP will need to extend the scope of coverage to poorer farmers in less favorable
agroecological zones.


6.1.1. Issues of Scale

Increases in the number of DAs (extension agents) have not kept pace with the rapid expansion in
the numbers of NEP participants. Extension experts in the Ministry of Agriculture recommend a











ratio of about one DA per 100 demonstration plots. However, it is estimated that this ratio now
ranges from one DA per 150 to as many as 500 demonstration plots in some areas. At this level,
the DAs may not be able to provide sufficient technical assistance to each participating farmer.
With the increased load comes additional responsibilities for credit administration (see Section
6.2) that also reduce the time agents can spend interacting with farmers on technical training.


6.1.2. Broadening the Scope of Coverage

The movement to less favorable agroecological zones and poorer farmers raises questions about
what will happen to yield response for the recommended technologies. The most logical
hypothesis is that the response will diminish and the production risks will increase as the program
reaches out to a broader population of farmers. Agents will increasingly be working with farmers
who need more supervision than the previous round of participants, but the agents will have less
time to devote to each one. The extension service has already begun to address this problem by
encouraging previous graduates and local school teachers to serve as volunteer assistants to the
extension agents during peak periods (planting, in particular), but this is a situation that will need
to be monitored closely as the program evolves.

There is not a sound basis for estimating the size of the yield gap or increased riskiness of the
technology as the adoption frontier moves out, but some preliminary results based on a CERES
modeling exercise suggest that the cumulative probability of yields less than 4 tons per hectare
increases from .05 to .25 when moving from the high rainfall zone of Jimma (1570 mm/year) to a
slightly lower rainfall zone such as Ambo (995 mm/year) (Schulthess and Ward 1999). In other
words, program expansion implies not only a challenge for the extension agents who will
progressively encounter farmers with less capacity to adopt the technologies but also for
researchers who must work to adapt improved varieties and other technologies to more difficult
production environments.


6.1.3. Cost Issues

Keeping the costs of the extension program under control is another real challenge. To date,
program expansion has been achieved largely through increases in the number of demonstration
plots supervised by individual agents, but there are limits to how far this can go and legitimate
concerns about the probable costs in terms of decreased quality of services. The NEP's failure to
"wean" farmers from the program after two years is another factor pushing up costs as it limits
the number of farmers trained and raises the costs per farmer. Weakness in both credit allocation
(Section 6.2) and input market development (Section 6.3) are making it difficult for farmers to
"graduate" from the program within the prescribed two-year period.











6.2. Improving the Credit System


6.2.1. Evolution of the "Regular" Credit System

In the early 1990s the agricultural input credit system in Ethiopia was in a state of disarray with
credit recovery rates declining from 54% in 1990 to 37% in 1991 and only 15% in 1992. By
1994, a new system was in place. Two government banks, the DBE and the CBE, began to
provide credit to farmers' organizations (i.e., not to individuals). The DBE and CBE have a total
of only 185 branches throughout the country-about one branch for 44,000 households. The
credit is guaranteed by regional governments, who play an important role in the administration of
the credit system (estimating input demand and credit needs, certifying creditworthy farmers'
organizations, etc.). In some regions (Amhara, Tigray, and Southern) the government actually
gathers loan applications, processes them, and issues purchase orders to suppliers-leaving the
banks with the relatively easy task of disbursing payments to designated suppliers. In other
regions (primarily Oromiya) the banks deal directly with farmer organizations that the government
has certified to be creditworthy. Since 1994 the credit system has shown some signs of
improvement: loan disbursements increased from 187 million birr in 1994 to 471 million approved
for disbursement in 1996 and repayment rates have been much better than in the early 1990s. At
the same time, a number of problems exist that raise questions about the ability of the system to
respond efficiently to farmers' growing demand for input credit.

The most common critique of the existing system concerns the impact that it has had on the
development of a private sector fertilizer marketing network. The administrative rules for
allocation of credit have led to defacto input distribution monopolies in several regions because
the regional governments tend to issue delivery orders to only one firm. These monopolies are in
the hands of the major fertilizer importers/wholesalers-sometimes they are companies owned by
the regional government (i.e., mini-parastatals). This system leaves little room for the
development of a competitive retail sector (see Section 6.3). It is within this context of change in
agricultural credit that the NEP introduced a second line of input credit tied to the extension
program.


6.2.2. The Beginning ofSG2000 and NEP Credit

Credit for the early MOA/SG participants did not pass through the official input credit program
described above but was made available through a parallel system guaranteed by SG2000 and
administered by the extension personnel assigned to the SG program. Early SG participants made
a 50% down payment to obtain inputs with the balance due at harvest. No interest was paid
because SG2000 believes that its nonprofit status prohibits this. Reimbursement rates were better
than 95%. When the NEP began in 1995, credit was made available to participants by setting up a
line of "extension" credit which runs parallel to the line of "regular" credit described above.











As in the regular credit program, extension credit is guaranteed by the regional governments and
administered jointly by them and the two government banks (DBE and CBE).The rules for NEP
credit are standard but differ from the earlier SG2000 rules: the down payment required is only
25%20 and interest is charged at prevailing rates (15% until 1996 when, following a reduction in
inflation, it was reduced to 10.5%). Unlike regular credit, which is issued only to organizations,
extension credit is allocated to individuals who sign up for participation in the NEP demonstration
plot program with their local extension agents (DAs).

Hidden But High Administrative Costs: The DAs, as well as other government personnel at the
regional, zone, and wereda levels, are heavily involved in the administration of the extension
credit-signing up individuals, determining their creditworthiness, collecting down payments,
issuing delivery orders and/or organizing public bidding procedures, and collecting payments after
harvest. Many of the DAs interviewed in August 1998 expressed concern about their heavy
involvement in credit administration-not only because it kept them from their technical
responsibilities (advising farmers on use of improved technologies), but also because their role as
credit collection agents had a negative impact on their personal relationships with farmers. The
current division of tasks among farmers, banks, and government appears to increase the public
cost of the credit program because many of the tasks performed by government personnel are
ones that would normally be performed by farmer organizations or banking sector personnel.
Creative solutions are needed to alleviate these pressures and some are already being tested. For
example, in one wereda, the DA has organized a group of farmers who are now assisting him with
the collection of down payments and reimbursements.

Competition Between Extension and Regular Credit: Another problem that became evident in
1998 is the strong competition between the regular and the extension program for the limited, and
sometimes declining, portfolio of available credit. The Oromiya region provides an illustration of
increasing problems with credit availability. Although the aggregate amount of agricultural credit
has increased fairly consistently from 1993/94 through 1997/98 the allocation by zone has been
irregular, leading to sharp reductions for some zones and important increases for others (Table
25). This pattern is inconsistent with the NEP program objectives of introducing farmers to new
technologies that they can continue to use after program assistance stops, as few farmers appear
to have the financial capacity to pay cash for the full technology package.









20 When the NEP instituted a 25% down payment, SG2000 also reduced theirs to the same level, but they
maintained their policy of no interest.











Table 25. Fertilizer Credit in the Three MOA/SG Study Zones, Oromiya Region
('000 Ethionian Birr)


Zone 1993/94 1994/95 1995/96 1996/97 1997/98
Jimma 6,753.0 17,432.4 16,426.0 14,080.5 13,246.4
East Shoa 39,417.5 55,201.3 62,705.8 52,663.6 40,559.0
West Shoa 27,861.3 44,633.1 57,002.9 42,159.4 51,280.0
Oromiya Region 138,295.2 190,479.2 211,135.6 210,758.6 225,971.6
Total*


Source: Oromiya Regional Government, Addis Ababa, Ethiopia, 1998.
* All zones, not just the three listed above.


Surveys in August 1998 revealed that the problem of competition between extension and regular
credit was important. In several zones and weredas the entire credit allotment was used for the
NEP, leaving nothing for nonparticipants accustomed to getting credit through their cooperatives.
The underlying reason appears to be the limited amount of credit allocated by the regional
government, coupled with a desire to meet NEP targets. For example, only one of the three
weredas surveyed in the Jimma zone extended fertilizer credit to farmers outside of the NEP in
1998. This means that SG/NEP graduates in the other two weredas will have to participate in the
program again to obtain fertilizer credit. It is reported that the DAs enroll graduate farmers
readily because they are known to repay their loans. However, this practice reduces the chance
that a farmer who has not participated in the program will be chosen and limits the amount of
regular credit available for nonparticipants. If this phenomenon continues, it risks seriously
compromising the ability of NEP participants to become true graduates capable of purchasing
improved technologies on their own.

Reimbursement remains an important issue for the viability of the credit program. Although
overall rates remain acceptable, they are lower than those obtained by SG2000 and will probably
decline as the NEP moves to poorer farmers and less favorable agroecological zones. As regional
governments are guaranteeing these loans, higher defaults will increase government costs and
reduce the economic profitability of the program. In an effort to minimize their losses, regional
governments have frequently called on the local police to enforce repayment. This has resulted in
some confiscation or forced liquidation of farm assets (animals, equipment) to meet payments.
While farmers must repay if the credit system is to remain viable, liquidation of assets must be the
exception rather than the standard procedure for ensuring reimbursement. As the program moves
to more risky farming situations, some type of renegotiated payment schedules should be
considered to allow for truly poor harvests.











6.3. Developing a Transparent, Responsive, Low-cost Input Supply System

The government's stated goal for fertilizer sector development is to promote a streamlined,
competitive and efficient fertilizer importation and marketing system to ensure the availability of
fertilizer on a sustainable basis (National Fertilizer Policy 1993, cited in World Bank 1995).
Unfortunately, the government's concurrent objective to decentralize many administrative
processes has led to very uneven progress toward fertilizer subsector goals. Regional
governments have become major actors in fertilizer markets, designing the rules and regulations
that shape fertilizer distribution within their region and, in several cases, creating what amounts to
regional parastatals with monopoly control of fertilizer marketing activities. Although substantial
progress has been made in liberalizing fertilizer imports, much remains to be done to improve
internal fertilizer distribution systems, reduce overall costs, and respond to farmers' overall needs
for both fertilizer and improved seeds.


6.3.1. Encouraging Transparency and Competition in the Fertilizer Sector

Fertilizer Imports: The current process of importing fertilizer is largely a function of procedures
established by the World Bank-funded Ethiopia National Fertilizer Sector Project. The project
encouraged the establishment of the National Fertilizer Industry Agency (NFIA) which has the
general task of assisting with and monitoring the liberalization of the fertilizer market. The NFIA
helps organize a series of foreign exchange auctions devoted exclusively to fertilizer (all other
foreign exchange in Ethiopia is made available through a general auction covering all other
goods). Any licensed importer (currently five in number) can submit a bid. Most fertilizer
auctions provide funds for imports of 25-30,000 tons (the maximum size of ship that can be
handled at ports serving Ethiopia) but occasionally bids are for smaller quantities (e.g., 10,000
tons-a less economical size of shipment).

Under the terms of the World Bank project agreement, foreign exchange is provided by the GOE
and/or bilateral donors for the first 160,000 metric tons per year as well as for 33% of imports
above this base amount. The International Development Association (IDA) of the World Bank
provides the project with foreign exchange for 67% of the imports over the base tonnage (this
import support represents about 92% of the $191.1 million three-year project portfolio).

Importers interviewed in August 1998 identified several characteristics of the bidding system that
tended to increase landed costs of fertilizer, but most acknowledged that many of the
shortcomings experienced the first year (particularly very long delays between bidding and
receiving foreign exchange) have diminished in importance. Among the most frequent complaints
are:











S Timing and infrequency of the auctions does not always permit importers to take advantage
of seasonal dips in fertilizer prices (foreign exchange auctions for other products are held
weekly);

S Conditionalities on bilateral funding often slow down the process and increase costs (e.g.,
restrictions on suppliers or shippers);

S The need to obtain letters of credit from banks for the full amount of the foreign exchange is
often slow and the amount of collateral demanded (generally 100%) is considered excessive
by some importers;

The length of time that passes from the moment that the importer gets a cost estimate from
his supplier and the time that he actually has the foreign exchange to finalize the transaction
is so long (one to two months) that suppliers add substantial margins into their estimates to
protect them against unexpected price fluctuations during the interval; and

S Some importers believe they could obtain foreign exchange and credit from their suppliers at
better rates than what is available locally, but such transactions are currently illegal.

Despite these problems, fertilizer imports have followed a general upward trend since 1993,
though with important inter-annual fluctuations due to large carry-over stocks in some years.
Table 26 summarizes data on aggregate imports and shares of the market going to each importer.
Though the playing field at the distribution level (see below) remains quite uneven, competition
exists at the import level and there is evidence of new players entering the market (e.g., Fertline
and Guna in 1998).











Tahip 7C; FPrtuluzPr Imnort Trends: 1993 to 1998


Year AISE EAL Fertline Ambassel Guna Total
Metric Tons of Fertilizer Imported

1993 not avail, not avail, not avail, not avail, not avail. 175,000

1994 not avail, not avail, not avail, not avail, not avail. 68,200

1995 232,219 55,400 0 0 0 287,619

1996 219,574 94,669 0 24,337 0 349,580

1997 160,000 0 0 0 0 160,000

1998 179,808 56,000 35,000 61,000 50,000 381,808


Source: Compiled from National Fertilizer Industry Agency, 1998.
Notes: AISE is a government parastatal that formerly had a national monopoly on input supply. It now competes
with other importers but also maintains buffer stocks in case other suppliers fail to cover the market adequately.
EAL and Fertline are long-established private sector firms that recently entered the fertilizer import market (in
1995 and 1998, respectively). Ambassel and Guna are regional parastatals based in Amhara and Tigray,
respectively.


Fertilizer Distribution: As noted above, efforts to decentralize decision making to the regional
level have made it difficult to fully liberalize fertilizer markets. In Amhara, Tigray, and Southern
regions (three of the four principal fertilizer consuming regions) the market is dominated by a
single firm owned in whole or in part by the regional government. These firms are generally
selected by the regional, zone, or wereda governments as the principal suppliers for fertilizer made
available through both the NEP and the regular credit programs. Given the role that the
government plays in allocating credit (see Section 6.2 above), it is easy for government agents to
control which firm receives the supply orders for fertilizer purchased on credit. The lack of
competition and transparency goes even further, however, as surveys revealed several instances of
private sector distributors having been denied the right to make cash sales from retail outlets in
these regions. The truly private sector firms (EAL and Fertline) as well as the national parastatal
(AISE) tend to become involved in these three regions only when the regional parastatal
experiences a shortfall. At such times help is solicited, usually from AISE first and then from the
others.

In the fourth region (Oromiya), a government-owned firm exists but the regional government
opted for a system of open bidding in 1998. The government firm (Dinsho) competed with other
distributors in a bidding process that awarded supply contracts for fertilizer purchased with NEP
and regular credit to the lowest bidder, providing they could show proof of stocks required to fill










the bid. It is interesting to note that it is in the Oromiya region where the government is also less
directly involved in administration of input credit-the wereda agricultural office certifies the
creditworthiness of farmers' associations, but representatives of these associations deal directly
with the banks to work out the details of the loans and make the payments after harvest.

Preliminary results of price analysis studies suggest that the Oromiya bidding system provides
farmers with lower cost fertilizer (after controlling for transport and other cost factors) than the
systems in the other three regions (Stepanek forthcoming). Nevertheless, many problems that
exist in the other regions also exist in Oromiya. One key issue is the lack of differentiation of
participants across functional areas. Importers are performing the full range of supply functions,
importing as well as covering most of the wholesale and retail transactions. This leaves very little
of the market for the large number of independent wholesalers and retailers who have been trained
by the NFIA with the objective of increasing competition in local markets. Further complicating
the picture is the fact that many service cooperatives tend to overestimate their members' demand
for credit purchases (i.e., they, in collaboration with wereda officials, overestimate the amount of
credit that is eventually allocated). This can result in substantial overstocks that the cooperatives
try to liquidate through cash sales, providing further competition for the independent retailers. It
remains unclear if the vertical integration of these activities and the manner in which the
cooperatives are assuming retail functions are in the interest of farmers (i.e., resulting in lower-
cost fertilizer) or not. This is an issue that requires further research.21

Policy Uncertainty Is A Problem for Importers and Distributors: In 1998 fertilizer importers and
distributors faced considerable policy uncertainty: stated fertilizer polices were often abandoned
and markets were not as open as the fertilizer distributors originally perceived. Given the
unpredictable policies across Ethiopia, fertilizer importers and distributors are unsure what
fertilizer policy will be enacted in the coming season. The national government has stated its goal
of developing a free market for fertilizer, but regional policies often carry different and changing
messages. Private wholesalers have frequently been unable to obtain a market share because the
regional government's policies favor certain importers and regional government companies.
These policy conflicts may be due to the government's efforts to decentralize administrative
responsibilities while simultaneously maintaining the right to set national goals. Policy
uncertainties at both the national and regional levels raise the cost of investing in the fertilizer
sector and may discourage new entrants.








21 For example, some cooperatives are adding margins to both their credit and cash sales that are greater than
margins charged by other distributors.











6.3.2. Reducing Costs


Input costs (especially fertilizer, but also seed) are a large component of the financial budgets for
farmers in the MOA/SG program (see Section 4). Reducing the cost of inputs will enhance the
accessibility of the program to a broader population and make the technology more profitable for
early adopters.

The preliminary results of the price analyses mentioned above suggest that fertilizer costs can be
reduced by increasing transparency and competition in the markets. There are a number of other
potential areas for cost reduction that were mentioned by the market participants interviewed.
Among the most common suggestions were:

* Better timing of imports (importing earlier to take advantage of seasonal drops in world
prices and completing inland distribution before roads deteriorate during the rainy season;

* Larger allotments of foreign exchange to obtain greater economies of size/scale in
importing;

* Improved infrastructure (options mentioned were better roads and the building of an
inland storage facility just inside the Ethiopian border to reduce excess port charges-
these options would benefit other importers and exporters as well);

* Helping farmers' organizations to take on more of the responsibilities for credit
administration and input delivery; and

Better estimates of demand to avoid the expense of carry-over stocks.22

There is a need to look more closely at the relative costs and benefits of these as well as other
options for reducing costs of fertilizer.


6.3.3. Meeting Needs ofAll Farmers

Fertilizer Credit: In 1998 the input supply system failed to meet the needs of farmers in several
different ways. In many cases, the problems stemmed from the manner in which the credit system
was managed. For example, many farmers not participating in the NEP program were unable to
get credit through the regular credit program and were therefore forced to cut back on fertilizer
use. The NEP appears to be absorbing more and more of the agricultural credit resources as it


22Demand is generally estimated by extension agents in collaboration with farmers' organizations, with the latter
tending to be overly optimistic about how much effective demand there really is.











expands, and the total amount of credit available for agriculture is not increasing proportionally.
Although the NEP has expanded, it is reported that in two of the three MOA/SG survey zones
total credit declined between 1997 and 1998. In Jimma the amount of fertilizer credit disbursed
fell by 5.9% between 1997 and 1998. In East Shoa the amount of fertilizer credit disbursed fell by
23% between 1997 and 1998.23 Where there were once multiple channels for credit accessible to
farmers, now in some areas NEP is the only credit source.

In some cases, non-participants who had cash still had problems because regional governments
were discouraging cash sales (a problem in Amhara, for example). This made it difficult
(sometimes impossible) for farmers to "graduate" from the NEP program because the only way to
get credit in many cases was to sign up again for a demonstration plot-this clearly limits the
capacity of the NEP to reach new farmers with their extension message. The shortage of regular
credit also made it difficult for farmers who had previously used fertilizer but never participated in
the NEP to find credit to maintain their past levels of fertilizer use.

Improved credit targeting and simultaneous development of the fertilizer cash market would
probably serve a broader group of farmers better than the current system. The better-off farmers
often join the NEP credit program simply to obtain improved seed that is not otherwise available
on the market (see below). Encouraging these farmers to purchase some fertilizer for cash or
through the regular credit program would free up NEP credit and extension resources that could
be targeted to poorer farmers.

Access to Improved Seed: Seed market development lags behind that of the fertilizer sector,
despite the urgent need to increase the availability of maize hybrids being promoted by the NEP.
Improved varieties of teff seed are available in local shops and markets, but hybrid maize seed is
not. Furthermore, there is no credit available for hybrid seed purchased outside of the NEP.
Apart from the National Seed Enterprise, a government parastatal that supplies the majority of
hybrid maize seed, the multinational firm Pioneer Hi-Bred is the only other major actor in the
Ethiopian maize seed industry. Pioneer usually sells its hybrids to the National Seed Enterprise,
but is now beginning to promote direct cash sales to farmers.

Both the credit and supply constraints of hybrids have important implications for NEP graduates
since without hybrid seed the responsiveness of fertilizer declines dramatically. Even graduate
farmers who are able to get fertilizer through the regular credit program end up participating in
the NEP again simply to have access to the hybrid maize seed.24



23 Credit information for 1997 and 1998 from unpublished data provided by the Commercial Bank of Ethiopia.

24 DAs anxious about filling their quotas for participants will often ignore the fact that an applicant has already
been in the program for two years.










REFERENCES


Abraham, Tadesse, Firdu Azerefagu, Assefa Gabre Amlak, and Adhanom Negassi. 1993.
Research Highlights on Maize Insect Pests and Their Management in Ethiopia. In
Proceedings of the First National Maize Workshop ofEthiopia, eds. Benti Tolessa and
Joel K. Ransom. Addis Ababa: IAR/CIMMYT.

FAOSTAT Statistics Database. 1999. Food and Agriculture Organization of the United Nations,
[Online]. Available: http://apps.fao.org/

Federal Democratic Republic of Ethiopia (FDRE). 1997. Agricultural Sample Survey 1996-97:
Vol. 1: Report on Area and Production for Major Crops. Statistical Bulletin 171. Addis
Ababa: Central Statistical Authority.

Gordon, Henry, Habtemariam Abate, and Kiflu Bedane. 1995. An Analysis ofFood Crop
Production, Productivity and Marketing in Ethiopia. Final report submitted to USAID
Ethiopia Mission by the Cropping System and Area Selection Team. Addis Ababa:
SAID.

Howard, Julie, Mulat Demeke, Valerie Kelly, Mywish Maredia, and Julie Stepanek. 1998. Can
the Momentum be Sustained? An Economic Analysis of the Ministry of
Agriculture/Sasakawa Global 2000's Experiment with Improved Cereals Technology in
Ethiopia. MSU Department of Agricultural Economics Staff Paper No. 98-25. East
Lansing: Michigan State University.

Kassahun, Aberu. 1998. Transport and Storage Analysis for Ethiopian Fertilizer Procurement and
Distribution. Addis Ababa, Ethiopia. Mimeo.

Sasakawa-Global 2000 Agriculture Project in Ethiopia. 1996. Annual Report Crop Season 1995.
Addis Ababa: SG2000.

Schulthess, Urs, and Richard Ward. Forthcoming. Using CERES-Maize to Assess the Impact of
SG2000 Technology on Maize Yield Stability in Ethiopia. East Lansing: Michigan State
University.

Seyfu, Ketema. 1993. Tef (Eragrostis tej): Breeding, Genetic Resources, Agronomy, Utilization
and Role in Ethiopian Agriculture. Addis Ababa: Institute of Agricultural Research.

Stepanek, Julie. 1999. Analysis of World DAP and Urea Fertilizer Prices 1995-98. East Lansing:
Michigan State University. Mimeo.











Stepanek, J. Forthcoming. How the Organization of the Fertilizer Sector in Ethiopia Affects
Maize Productivity. Ph.D. dissertation, Michigan State University.

World Bank. 1995. Staff Appraisal Report, Ethiopia, National Fertilizer Sector Project. Report
No. 13722-ET. Washington, D.C.: World Bank.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs