i.
. .‘1 p. a.
..2L«.u~

a!
.I 1.1,
(I A

. $5,
. u:

 

ll?
'lxl .3;
.\:l..

13":
1| . II.

xi i .i {It

.\

‘ .r..,

3.1.3:...
it 3.:

1.. .25.!
‘71!
laid—”l
‘8, . pi 4.
.2 1. 3.: tr
‘ 2!! . . ..: ”14...?“
$75. .5 .bMerxn

. Ilnkhu. ..h H , V
3. £9533...»
- \

 

llllll'llllllllll‘llilllllllll llllllllllll -

3 1293 01026 9763

This is to certify that the

thesis entitled

Ex Post Assessment of Investments in
Cameroon's Cowpea and Sorghum
Research-Extension Systems

presented by

James Arthur Sterns

has been accepted towards fulfillment
of the requirements for

 

 

Master of Science degree inAgricultural Economics

TRMVLEMTR

Major professor

Date—3J16!q3

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

LIBRARY
Michigan State
Universlty

 

 

 

PLACE IN RETURN BOX to roman this checkout from your record.
TO AVOID FINES Mum on or baton duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Msu I. An Affirmative Action/Equal Opportunity Institution
Wampum

 

EX POST ASSESSMENT OF INVESTMENTS IN CAMEROON’S
COWPEA AND SORGHUM RESEARCH-EXTENSION SYSTEMS

By

James Arthur Stems

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Agricultural Economics

1993

ABSTRACT

EX POST ASSESSMENT OF INVESTMENTS IN CAMEROON’S
COWPEA AND SORGHUM RESEARCH-EXTENSION SYSTEMS

By

James Arthur Sterns

Throughout Africa, per capita food production has been declining since the early 19603.
Cameroon has sought to counter this trend by increasing agricultural productivity
through research and extension. In order to establish future investment priorities, policy
makers need to know if past agricultural research investments have earned sufficient
returns to justify continued funding. To address this issue, data were collected in
Cameroon and analyzed in order to estimate the benefits and costs of investments in
sorghum and cowpea research and extension in northern Cameroon. Focusing on the
period 1979 to 1987, the analysis addressed three questions: what were the returns to
past investments, what factors explained the estimated returns, and how were the
benefits distributed. Results include estimated internal rates of return of 15% for
W research and extension, and 1% for sorghum research and extension. The
analysis also highlights the impact of institutions on the returns to investments and why
similar investments in parallel research programs within the same research-extension

system had diflerent returns.

Capyright by
JAMES ARTHUR STERNS

1993

This thesis is dedicated to my parents, John and Harriett Stems,

whose love I treasure.

ACKNOWLEDGMENTS

First in a long queue of friends and colleagues who deserve recognition for their
support, care, and nurturing are Noah, Colin and Carol Evans with whom I have shared
much random kindness and many senseless acts of beauty.

Next are the graduate students, support staff and faculty of the Department of
Agricultural Economies, Michigan State University. Kudos especially to Dr. Richard
Bernsten, for being a mentor in work and life. Thanks also to the other members of my
committee: Dr. Eric Crawford for unwittingly assuming the role of surrogate father
(only man I’ve ever met who could rival my dad’s skill at instilling guilt quite
unmaliciously and unintentionally) and Dr. Russ Freed, the first MSU faculty member I
ever met (while sharing a Beaufort in Jerry J ohnson’s front yard in Maroua, Cameroon).

The people of Cameroon, and especially the research staff at IRA-Maroua and
within the NCRE project, must be noted here. My life has been deeply enriched by the
times I have spent with them. May God bless their country and their work. In
particular, the guidance provided by Jerry Johnson, John Russell, Mulumba Kamuanga,
Doyle Baker and Lauri Kitch were sincerely appreciated. A heartfelt thanks is also
extended to Georges Dimithé and his family.

I also wish to acknowledge USAID, whose funding, through the Food Security in
Africa Cooperative Agreement, made this research possible.

Finally, my appreciation goes out to the parishioners and staff of All Saints
Episcopal Church. The Living Word is a wonderful reminder of what really matters.

And thanks to Jen, because she deserves it.

TABLE OF CONTENTS

LIST OF TABLES .................................................. viii
KEY TO ACRONYMS ................................................ x
CHAPTER 1 INTRODUCTION ........................................ 1
1.1 Problem Statement .......................................... 1

1.2 Objectives ................................................. 1

1.3 Hypotheses ................................................ 2

1.4 Organization of Thesis ....................................... 3
CHAPTER 2 THE PROBLEM SETTING ................................ 4
2.1 Overview of Cameroon ....................................... 4

2.1.1 Social-Political Evolution ............................... 4

2.1.2 Monetary Policy and Trade Status ........................ 6

2.2 Cameroon’s Agricultural Sector ................................ 8

2.2.1 Overview ........................................... 8

2.2.2 Northern Cropping Systems ............................. 8

2.2.3 Land in Crop Production .............................. 11

2.2.4 Rainfall ........................................... 15

2.3 Agricultural Research in northern Cameroon ..................... 16

23.1 Overview .......................................... 16

2.3.2 Sorghum and COWpea Research ......................... 17

2.3.3 Staffing and Financing ................................ 18

2.4 Agricultural Extension ...................................... 19

2.4.1 IRA’s Farming System Research Efforts .................. 19

2.4.2 MINAGRI’s Extension Efforts .......................... 20

2.4.3 SODECOTON Extension Efforts ........................ 20

CHAPTER 3 LITERATURE REVIEW AND METHODOLOGY ............. 23
3.1 Review of Literature ....................................... 23

3.1.1 Introduction ........................................ 23

3.1.2 Methodologies ...................................... 24

3.1.3 An Underlying Assumption ............................ 27

3.2 Benefit-Cost Approach ...................................... 28

3.2.1 Overview .......................................... 28

3.2.2 Advantages of Benefit-Cost Analysis ..................... 29

3.2.3 Potential Disadvantages of Benefit-Cost Analysis ............ 30

3.3 An Approach to Institutional Analysis .......................... 31

3.3.1 Overview .......................................... 31

vi

vii

3.3.2 The Contribution of Institutional Analysis ................. 32

3.4 Data Collection ........................................... 33
3.4.1 Data Availability .................................... 33

3.4.2 Data Base Compilation ............................... 34

3.4.3 Data Reliability ..................................... 35
CHAPTER 4 RATE OF RETURN ANALYSIS ........................... 36
4.1 General Approach ......................................... 36
4.2 Cowpea ................................................. 37
4.2.1 Overview of Improved Cowpea Technologies ............... 37

4.2.2 Cowpea Cost Stream ................................. 39

4.2.3 Cowpea Benefit Stream ............................... 51

4.2.4 Base Run for Cowpea ROR ............................ 70

4.2.5 Sensitivity Analysis - COWpea ........................... 75

4.3 Sorghum ................................................. 83
4.3.1 Overview of Sorghum Technologies ...................... 83

4.3.2 Sorghum Cost Stream ................................ 85

4.3.3 Sorghum Benefit Stream .............................. 96

4.3.4 Base Run for Sorghum ROR ........................... 106

4.3.5 Sensitivity Analysis - Sorghum ......................... .. 108
CHAPTER 5 INSTITUTIONAL ANALYSIS ............................. 117
5.1 Agricultural Development in Northern Cameroon ................ 117
5.1.1 Efforts Prior to 19708 - King Cotton .................... 118

5.1.2 Early 19708 - Era of Government Interventionism .......... 119

5.1.3 Early 19803 - The PCN Era ........................... 123

5.1.4 Late 19808 - Restructuring ........................... 124

5.2 Key Institutional Linkages, 1979-1987 ......................... 126
5.2.1 linkages within the local system ........................ 126

5.2.2 Linkages beyond the local system ........................ 128

5.2.3 Government Policies ................................. 129

5.3 Distribution of Benefits ..................................... 129
CHAPTER 6 SUMMARY AND CONCLUSIONS ......................... 132
6.1 Reported Rates of Return ................................... 132
6.1.1 A critique of IRR Estimates ........................... 132

6.1.2 Are these financial or economic returns? .................. 134

6.1.3 Qualitative Issues .................................... 136

6.2 Differences in the Rates of Return ............................. 137
6.3 Summary of Key Lessons .................................... 140
6.4 Institutionalizing Assessment Methodologies ..................... 141

BIBLIOGRAPHY ................................................... 146

Table 2.1
Table 2.2
Table 2.3
Table 2.4
Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table 4.5

Table 4.6
Table 4.7
Table 4.8

Table 4.9

Table 4.10

Table 4.11

Table 4.12

LIST OF TABLES

Macroeconomic Indicators and Trade Figures for Cameroon, 1980

to 1989. .................................................
Sorghum and Total Harvested Crop Hectares, Far North Province,
Cameroon, 1984 to 1989. ...................................
W and Total Harvested Crop Hectares, Far North Province,
Cameroon, 1984 to 1989. ...................................
Number of Senior Researchers, IRA Research Center, Maroua,
Cameroon, Selected Years. .................................
Extension Recommendations for Cowpea, Northern Cameroon,

1984 to 1991. ............................................
Estimated Total Expenditures (Nominal SUS), Cowpea Research,
SAFGRAD Joint Project No. 31, Maroua, Cameroon, 1979 to

1987. ..................................................
Estimated Total Expenditures (Nominal SUS), Bean/Cowpea
Collaborative Research Support Project, Maroua, Cameroon, 1982

to 1987 . ................................................
Estimated Total Expenditures in Nominal SUS for the IRA-

Cowpea Program, Maroua, Cameroon, 1982-1987, Costs Estimates
from Two Sources. .......................................
Estimated Total Expenditures (Nominal SUS), SODECOTON
Extension Costs for Cowpea Technologies, Far North Province,
Cameroon, 1981 to 1987. ...................................
Estimated Total Costs (Nominal SUS) for Cowpea Research and
Extension Programs, Far North Province, Cameroon, 1979 to 1987. . .
Cowpea Yield Estimates, Various Sources, Far North Province,
Cameroon, 1983 to 1990. ...................................
Sorghum Yield Data, Various Sources, Northern Cameroon, 1983

to 1990. ................................................
Projected Adoption Rates for COWpea Technologies, Far North
Province, Cameroon, Years 1984-1998. ........................
Input Prices for COWpea Production and Costs Associated with
Adopting the Improved Cowpea Package, Far North Province,
Cameroon, 1984 to 1998. ...................................
Estimated Average Annual Market Prices for Cowpea and

Sorghum By-Products, Far North Province, Cameroon, 1984-1998 . . .
Gross Benefits, in ’000 $US, from the Development and Extension

of Improved Cowpea Technologies, Far North Province,

Cameroon, 1984 to 1998. ...................................

viii

7

13

15

18

39

43

44

49

50

52

58

64

67

70

73

Table 4.13

Table 4.14

Table 4.15

Table 4.16

Table 4. 17

Table 4.18

Table 4.19
Table 4.20

Table 4.21

Table 4.22

Table A.1

Table A2

Estimated Benefit-Cost Flows (in ’000 $US) for the Improved

Cowpea Package, Far North Province, Cameroon, 1979 to 1998. ..... 74
Sensitivity Analysis, Modifying Values of Key Variables and

Subsequent Changes in the ROR for Cowpea Research &

Extension, Far North Province, Cameroon, 1979-1998. ............ 76
Estimated Total Expenditures (Nominal SUS), Sorghum Research,
SAFGRAD Joint Project No. 31, Maroua, Cameroon, 1979 to

1986. .................................................. 87
Estimated Total Expenditures (Nominal SUS), Sorghum Research

within the National Cereals Research and Extension Project,

Northern Cameroon, 1982 to 1986. ........................... 90
Estimated Total Expenditures by IRA, in Nominal $US, for

Sorghum Research at IRA’s Maroua Research Center, Cameroon,

1979 to 1986. ............................................ 93
Estimated Total Expenditures, Nominal SUS, SODECOTON

Extension Costs for Sorghum Technologies, Northern Cameroon,

1981 to 1986. ............................................ 94
Estimated Total Costs, Nominal SUS, Sorghum Research and

Extension Programs, Northern Cameroon, 1979 to 1986 ............ 95
Projected Adoption Rates and Estimated Area Harvested for $35 .
Sorghum, Northern Cameroon, Years 1984-1998. ................ 104

Estimated Benefit-Cost Flows (in ’000 $US) for the Development

and Extension of the Improved Sorghum Variety S35, Northern

Cameroon, 1979 to 1998. ................................... 109
Sensitivity Analysis, Modifying Values of Key Variables and

Subsequent Changes in the ROR for Sorghum Research &

Extension, Northern Cameroon, 1979-1998 ...................... 110
Sorghum, Hectares Harvested, Far North Province, Cameroon for
the Years 1972/73 to 1989/90. .............................. 144

Hectares Harvested, Far North Province, Cameroon for
Years 1972/73 to 1989/90. ................................. 145

ARR

CRSP
CCCE
CFDT

DEAPA

EEC
FAC
FAO

fcfa
FONADER
GDP
IARC
ICRISAT
IITA
IRA
IRAT

IRCI'
IRR

IMF
MESIRES

MIDEVIV

MINAGRI
MRR
NAETP
NCRE
NCSM
NEB

N GO
ONAREST
PCN

PI

ROR

SAF GRAD
SEMRY

SODECOTON
TLU
USAID

KEY TO ACRONYMS

Average Rate of Return

Collaborative Research Support Project

Cain-e CentmIe de Cooperation Economique

la Campagne ancaire pour le Developpement des Fibres
Textiles

Division des Enquetes Agroéconomiques et de la Planification
Agricola

European Economic Community

Fonds d {Aide et Cooperation

Food and Agricultural Organization, United Nations

fiunc de la Communauté Financiere Afi'icaine

Fonds National de Développement Rural

Gross Domestic Product

International Agricultural Research Center

International Center for Research in Semi-Arid Tropics
International Institute of Tropical Agriculture

I ’Institut de Recherches Agronomiques

I ’1me de Recherches Agronorniques Trcpicales et des
Cultures Vivrieres

I ’Institut de Recherche sur le Coton et Fibres Textiles
Internal Rate of Return

International Monetary Fund

Ministere de I ’Enseignement Superieur, de I ’Informatique et de
la Recherche Scientifique

Mission de Developpement des Cultures Vivrieres, Mamichéres
et Fruitiers

Ministry of Agriculture, Government of Cameroon
Marginal Rate of Return

National Agriculture Extension and Training Project
National Cereals Research and Extension Project

North Cameroon Seed Multiplication Project

Projet Nerd-Est Benoue

Non-Government Organization

Office National de la Recherche Scientifique et Technique
ijet Centre-Nora

Principal Investigator

Rate of Return

Semi-Arid Food Grain Research and Development project
leSociétéd’ExpansionetdeModemisationdelaRizicultured
Yagoua

Société de Développement du Coton

Testing-Liaison Unit

United States Agency for International Development

CHAPTER 1 INTRODUCTION

1.1 Problem Statement

Since the early 1960s, developing countries, assisted by foreign donors, have
invested resources to strengthen their agricultural research systems. Agricultural
economists have supported this strategy, arguing that technological innovations in
agricultural production drive the development of the agricultural sector, which in turn
contributes to the development of the general economy (Mellor, 1966; Eicher and Staatz,
1984). While several studies report a high rate of return to agricultural research in the
US, Europe, Asia and Latin America, there is considerable debate as to whether these
investments have netted positive returns in Sub-Saharan Africa (Oehmke, et al., 1991).

This quandary suggests that research is needed to address two critical issues.
First, in today’s world of limited resources and tight budgets, there is a need to
determine if past investments in technology-generating agricultural research in Sub-
Saharan Africa have generated sufficient returns to justify continued investments.
Second, there exists a need to examine national experiences in implementing agricultural
research in order to identify factors that explain the variability in the impact of these
investments.
1.2 Objectives

Cameroon, like many other countries, has sought to increase agricultural
productivity through research and extension of locally developed and / or screened
technologies. The general objective of this study is to assess the impact of the
development and extension of improved W and sorghum technologies in northern

Cameroon, and to describe factors that contributed to the observed impact.

2

The specific objectives of this study are to:

a.

Describe the evolution of the research and extension system since its
inception in order to provide a detailed background for the analysis;
Estimate the economic rates of return to sorghum and cowpea research
and extension in northern Cameroon, using a benefit-cost approach;
Review the institutional factors, linkages, and characteristics associated
with the research—extension system in order to determine how each
interacted to complement and/ or impede the performance of the W
and sorghum subsectors;

Discuss lessons learned from this study, focusing on how and why the
returns to research and extension differed between commodities;
Provide project managers and researchers with guidelines as to data
needed to more accurately monitor the impact of future agricultural

research and extension projects.

1.3 Hypotheses

This study tests the hypothesis that investments in agricultural research and

extension systems have positive returns. Specifically, it will determine if the internal rate

of return (IRR) of investments in com and sorghum research and extension in

Cameroon is greater than zero.

Qualitative analysis complements the quantitative analysis by identifying the role

of institutions and their influence on the productivity of agricultural research

investments. The hypothesis to be tested is that the necessary institutions and inter-

institutional linkages existed in northern Cameroon and that these institutions

contributed to a positive IRR.

1.4 Organization of Thesis

The thesis is divided into six chapters. Chapter 2 outlines the analytical
framework of the study, including general descriptions of Cameroon’s economy, and the
roles of its agricultural sector and agricultural research system. Chapter 3 briefly reviews
the rate of return literature, and presents the research design and methodology.
Chapter 4 reports results of the rate of return analysis. Chapter 5 identifies the
institutional factors that have contributed to the rate of return. Chapter 6 summarizes
the findings of the study and draws policy implications for institutionalizing impact

assessment.

CHAPTER 2 THE PROBLEM SETTING

2.1 Overview of Cameroon
2.1.1 Social-Political Evolution

Cameroon is a west-central African country bordered by six countries: Nigeria,
Chad, Central African Republic, Congo, Gabon, and Equatorial Guinea (Figure 2.1). Its
northern provinces have witnessed the rise and fall of various Sudanic civilizations for
the past 2,000 years, and its southern provinces are the possible origin of the Bantu
tribes some 2,500 years ago (Mbuagbaw; et. al., 1987). The Portuguese, the first
Europeans to visit the area, arrived in the late 1400s, but it was not until the late 18008
that the region as a whole was colonized by the Germans (although most activity was
concentrated in the south). After World War 1, France and Great Britain divided the
territory and colonized their respective shares.

French Cameroon gained independence in 1960, and British Cameroon became
independent in 1961. The two northern provinces of British Cameroon merged with
Nigeria while the southern province merged with the former French colony to form the
Federal Republic of Cameroon.

Today, the Republic of Cameroon is officially a bilingual country (French and
English) governed by an unitary republic. The country is divided into ten administrative
provinces, two of which were formerly under British rule while the remaining eight were
under the French. The 1976 census estimated the population to be 7.66 million people,
and the 1987 census estimate was 10.49 million, indicating an annual population growth
rate of 2.9% (National Directorate of the 2nd General Population and Housing Census,
1990). The 1987 census estimated that roughly 80% of the population were living in the

French-speaking provinces.

2.1.2 Monetary Policy and Trade Status
Cameroon’s currency is the franc de la Communauté Financiere Afi'icaine (fcfa),

exchanged at a fixed rate of 50 fcfa per French franc (1992). Economists familiar with
the fcfa zone of West-Central Africa generally agree that Cameroon’s currency is
currently overvalued by about 40%‘. Overvaluation reduces the competitiveness of the
country’s exports and underprices the real cost of imports.

Cameroon’s principal trading partners are France, the United States, the
Netherlands, Japan and Germany. Crude oil, coffee, cocoa, timber and cotton are the
country’s principal exports. Industrial equipment, semi-finished goods, food, drink,
tobacco, and vehicles are the principal imports. Over the period 1980 to 1989, the
market value of annual exports averaged $1.75 billion while the market value of annual
imports averaged $1.34 billion.

The economy experienced a period of rapid growth and overspending in the late
19703 and early 19803, fueled by the discovery of oil off Cameroon’s coast. For example,
real GDP growth reached 12.6% in 1980 and 13.3% in 1981. The general economy
declined after 1986, primarily due to drastic falls in world prices for all of Cameroon’s
exports. Restructuring of the economy and rescheduling of the external debt began in
1987, but the government has not yet fully implemented IMF austerity measures (1991).
Table 2.1 summarizes major macroeconomic indicators and trade figures for the past ten

years.

 

lEstimate based on interviews with personnel from the World Bank and
USAID / Cameroon. Salinger and Stryker (1991) found similar over-valuation (50%) in
other West African fcfa countries (Mali, Senegal, and Ivory Coast). According to these
authors, estimates of overvaluation are calculated by either comparing purchasing power
between trading partners, or by adjusting the exchange rate for unsustainable external
imbalance of payments and trade distortions.

38:58.. 298858 3. my: use £33 £5 assess 02808 nausea

5.03.332 605...: one .853 038%.: .55 new .5. 63256 .0: Sun .2: 8.8%.: .32..
A as? Ewen Eco» no.5 e A 39900 Sues when» n30 A... .auuacfium 3 .8 05:. ufiueo «30> A.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

od—m Siam mdom n63” new». 23. Sam 93*” EKN m; a anFCow .89. encasem—
mE hm: wdfi On: «.3 0.3 v.9 ”.2. Wen mdm 2.8 =E .ufioa .8on
an: 9mm— cdfl ofifi a: Wofi .33 a: a: a: ace“ :8 .96th «800
a: SUB o6: Ema S— we: ado a: a: a: nee“ 5: woo:— 3on
need and mmd cad 3...... 86 Sun a: a: s: 2.8 =E .605 we 09:0
§ .033

v.3 odm odm awn hum a: a: a: a: a: 315% Boo autoim—
acNm- can- 2.2- So- 5mm «mm- «mm. mAmm- 2.3.. Nomm- :E 33 £583 “cacao
awn; av.” an." we; ac.“ CA N; o: 3; 3.“, as $9 .50 3.59:—
gen.“ 3.“ K.“ cod end an.“ NBA 34 mm." 3.4 .5 3D 58 885
god. ofi N: v.3 O: W: 5.2 TS ed E. 3:22qu vote 5:528
.66. 9w- wd- ad I. 95 ad fie 02 v.3 as .25on mQO 18m
.5 £8

38mm ween «Now we; mean mam wGN mSN can“ 32 ..mootm goo—SE .mn—O
$3 $3 33 one 32 $3 30— $3 $3 83 398%:— omfioeeooobaz

 

 

 

 

 

 

 

 

 

 

 

 

$3" 8 83 6080an new ooh—BE coach. was 9.88%.: ciao—803.82

«N 035.

2.2 Cameroon’s Agricultural Sector
2.2.1 Overview

Cameroon’s agricultural sector is highly diverse, due in part to the wide range of
ecological zones found within the country’s borders. These zones, and their areas of
crop production as a percentage of the national land base, include highlands (27%),
savanna (22%), semi-arid plains (19%), equatorial rain forests (18%), and coastal
lowlands (14%); as shown in Figure 2.22.

This study focuses on the northern region of the country, which ranges from a
wooded, Guinea savanna in southern Adamaoua Province, to Sudanian and Sudan
Sahalian savannas in northern Adamaoua Province, all of the North province and much
of the Far North province, to the Sahel region of the Lake Chad area.

2.2.2 Northern Cropping Systems A

The three northern provinces are generally subdivided by principal cropping

systems and the underlying annual rainfall, which declines from south to north

(Figure 2.3).

 

2Estimated from data on shares (by province) of total land planted as reported in the
1984 Agricultural Census (National Directorate of the Agricultural Census, Yaounde).
The Southwest and Littoral Provinces were classified as coastal lowlands; the South, 60%
of the East, and 50% of the Center were classified as rain forests; the West and
Northwest were classified as equatorial highlands; 40% of the East, 50% of the Center,
the Adamaoua, and the North were classified as savanna; and, the Far North was
classified as semi-arid.

 

 

 

 

 

 

 

 

 

Figure 2.3 Rainfall isohyets.

10

W

Southern Adamaoua is a sparsely-populated area where maize is the principal
cash crop, and maize and tubers are the principal sources of food. Although this area,
with an annual average rainfall of 1750 mm, has great agronomic potential, tsetse fly has
historically been a constraint to production.
MW

Northern Adamaoua and nearly all of the North Province, with an average
annual rainfall of 1100 mm, is best described as the cotton and maize belt. Cotton was
first grown north of this zone, but northern Adamaoua now leads the country in cotton
production, primarily because the cotton parastatal has shifted its efforts southward into
areas of higher and less variable rainfall. Maize has traditionally been a
garden/compound crop in this zone, but since the mid-19803, maize has evolved into an
important cash crop. This development has been partially due to the creation of
MAISCAM, a private sector maize oil processing plant in NGaoundéré, the provincial
capital of the Adamaoua Province.
W

The Center-North zone3~including the Mayo Louti Department of the North
Province and the bulk of the Far North Province south of Wazanis the core of the
cotton-sorghum belt. With an average annual rainfall of only 800 mm, this area is also
plagued by erratic and uneven rainfall. Both low total annual rainfall and poor rainfall
distribution constrain production and often lead to drought-like conditions during the

growing season.

 

3This area is referred to as the Center-North zone because of a World Bank rural
development project by the same name that targeted this area over the period 1982 to
1987.

11

W

This case study focusses on the low rainfall Center-North zone where the
principal cropping patterns are either a cotton-sorghum two-year rotation, or a cotton-
sorghum-legume three-year rotation. Frequent variations within this general pattern
include the intercropping of sorghum with legumes, particularly cowPea but also
groundnut and bambara groundnut, the planting of a second crop of sorghum‘ late in
the growing season, and the substitution of pearl millet for sorghum in the production
cycle. In the Mandara Mountains, an area in the Far North Province but outside the
Center-North zone, farmers practice a biennial crop rotation-«planting sorghum one year .
and then intercropping pearl millet and a legume the next year. This rotation has
evolved as a strategy for controlling weeds and pests, especially the parasitic weed striga
(Striga Hermonthica). V
2.2.3 Land in Crop Production

Two data sources report harvested area for Cameroon. First, the Ministry of
Agriculture (MINAGRI) reported sorghum and cowpea production and area harvested
in northern Cameroon for the period 1972/73 to 1989/ 90. These time series indicate
large year-to-year fluctuations and no discernable trend (Appendix, Tables A.1 and A2).
Many key informants interviewed in Cameroon cautioned that these data were unreliable
because MINAGRI has historically had limited resources for data collection and
compilation.

Recognizing the need for more reliable data, the National Directorate of the

Agricultural Census (DEAPA), with support from USAID, initiated a project in 1984

 

‘Second crop sorghum in northern Cameroon, commonly called 'mouskwari", is
seeded in small, often irrigated, garden plots and then transplanted to the field late in
the growing season. It then matures on residual soil moisture.

12
with the explicit goal of estimating crop area and yields, based on farmer surveys and in-
field measurements. Researchers and other in-country agriculturalists consider these
data to be the best available. Yet, given the large year-to-year fluctuatiOns in estimated
harvested area and the short (six years) length of the time series, it is impossible to
discern historic trends or to project future levels of production and land use from these
data.
mum

Sorghum, and to a lesser degree pearl millet, are the traditional cereal gains of
the region and the primary sources of calories. In an average year, sorghum comprises
approximately 70% of total land harvested (Table 2.2). While MINAGRI and DEAPA
estimates of area planted to sorghum differ considerably in a given year, both data series
show that sorghum production is the most important food crop in the Far North
Province. For example, over the 1984-1989 period, DEAPA reports’, on average, an
estimated 332,000 ha in sorghum (73% of cropped area) while MINAGRI reports an
estimated 313,000 ha (54% of cropped area).

As reported in Table 2.2, two craps of sorghum are gown in northern
Cameroon. Dry season sorghum (mouskwari) is planted on vertisols late in the growing
season, maturing on residual soil moisture. Rainy season sorghum varieties are more
heterogeneous relative to mouskwari, difl‘ering by a geater degree in stalk length, gain

color (usually red or white) and length of gowing cycle (short, medium, or long).

 

’Census data estimates combined sorghum and millet data. Hence, to estimate the
sorghum area, the reported data were multiplied by 0.9 to remove pearl millet and the
adjusted data are reported. This factor is based on the judgments of key informants
involved in agicultural research in northern Cameroon.

13

Farmers’ preferences reflect this heterogeneity, with each farmer choosing varieties

which have specific traits that he or she desires.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.2 Sorghum and Total Harvested Crop Hectares, Far North Province,
Cameroon, 1984 to 1989.
Grain Sorghum, (Ha) All Crops
Year Rainy Dry Season Total Total Sorghum
Season (Ha) share (%)
1984 135,902 119,502 255,404 383,983 66.5
1985 185,424 135,030 320,454 445,380 72.0
1986 178,777 201,531 380,308 511,352 74.4
1987 95,676 147,856 243,532 335,154 72.7
1988 161,138 224,056 385,194 > 511,299 75.3
1989 172,313 232,116 404,429 519,143 77.9
mean 155,000 177,000 332,000 451,000 73.1
Source: National Directorate of the Agricultural Census, Cameroon.

Although sorghum production dominates the agiculture sector in northern
Cameroon, farmers face a host of constraints. Russell notes the following examples:

poor and erratic rainfall, often disastrously distributed during the gowing season;
striga, which is increasing in importance as both soil fertility and the length of
fallow period decrease; labor constraints at the time of sowing and weeding,
which impede improvement in land preparation and weed control; and lack of
credit for yield-enhancing inputs such as animal traction, fertilizer, and pesticides
(1991, p. 8).

Other constraints include a variety of insects and endemic leaf diseases.

29m

Cowpea, the second crop on which this study focuses, accounts for an estimated

5% of the harvested area. Like sorghum, cowpea is a traditional food crop in northern

14
Cameroon. In 1979, Perez, on an International Institute of Tropical Agiculture (IITA)

plant exploration tour, collected 396 samples of cowpea while in Cameroon, noting “an
impressive wide variability“ in cowpea varieties gown by farmers.

Although a relatively minor crop in terms of hectares harvested, several studies
(Ta’Ama, 1984; Wolfson, 1989; Kitch, 1990) have found that W contributes
significantly to household food security in northern Cameroon. First, because cowpea
matures early, households are able to harvest leaves and geen pods during the 'hungy
season” (late June through August) when gain reserves from the previous harvest are
depleted and farmers have yet to harvest this year’s crops. Second, com is an
important source of protein, especially for the rural poor. Singh and Rachie estimate
that cowpea contains 23 to 30% protein, with variations in content due to varietal
differences and environmental factors. Third, as a drought-tolerant crop that matures in
60 to 80 days on as little as 300 mm of rain, cowpea reduces farmers’ exposure to risk.
Finally, cowpea hay (leaves and stems) is used by limited-resource farmers to feed their
livestock during the dry season and to earn cash through sales in local markets.

Time series data for area planted to mm are even less reliable than the
sorghum estimates. As with sorghum, DEAPA data cover only a six-year period and
MINAGRI estimates are considered unreliable. In addition, neither agency reports
production figures specifically for cowpea. For example, DEAPA data classify cowpea in
the general category of I'beans"--which includes common beans, kidney beans, and
cowpea. Similarly, the MINAGRI time series reports nine years of data for "han'cats

daliques', after which data are reported for "haricotr/niébé". Finally, since W is

 

‘Haricots doliques is a French horticultural term referring to plants of the Dolichos
species and / or Vigna species, which would include cowpea. Haricot/niébé translates to
beans/mpea.

generally intercropped, it is extremely difficult to accurately estimate yields, implying that

15

even the DEAPA data is not entirely reliable.

Since cowpea is the only 'bean' crop gown on a large scale in the Far North
Province, this study assumes that 100% of the quantities reported for this province are
for cowpea (Table 2.3 and Appendix A2). The large amount of year-to-year variability
reported in Table 2.3 may be attributable to weather, changes in farming practices,
and / or human error in data collection and compilation. Yet, these best available data
indicate an average annual harvested area (1984.1989) of 23,600 ha which accounted for,

on average, 5.3% of the area harvested.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.3 Cowpea and Total Harvested Crop Hectares, Far North Province,
Cameroon, 1984 to 1989.
Area Harvested (Ha) Cowpea Share
Year COWpea All Crops (%)
1984 23,470 383,983 6.1
1985 30,232 445,380 6.8
1986 24,109 511,352 4.7
1987 16,744 335,154 5.0
1988 29,975 511,299 5.9
1989 16,998 519,143 3.3
mean 23,600 451,000 5.3
Source: National Directorate of the Agricultural Census, Cameroon.
2.2.4 Rainfall

In northern Cameroon, virtually all sorghum and cowpea production is rainfed.

Rainfall distribution is monomodal, usually beginning in late May, peaking in August and

 

16

ending in late October. However, there is a great deal of variability in this distribution,
due to late and unpredictable onsets of the rainy season, and a highly erratic distribution
of rainfall between localities. Russell notes that:

Northern Cameroon has suffered from an extended drought episode, that is, a

period in which droughty years are more frequent than usual. Farmers tend to

think that the current drought episode, which has lasted more than a decade and

a half, is due primarily to decreasing rainfall. Nicholson (1986) cites several

factors that may have contributed to drought, including overgazing,

overcultivation, and removal of vegetation, but concludes that the fundamental

cause of the current drought is meteorological (1991, p. 2).

In setting the research agenda, agricultural scientists working in northern
Cameroon have sought to take into account rainfall patterns. Any evaluation of the
research system needs to include some measure of the effects of rainfall on the general
state of the agricultural sector of northern Cameroon, and of how rainfall influenced
research agendas, the selection of technologies for extension, and ultimately, the returns

to research and extension efforts.

2.3 Agicultural Research in northern Cameroon
2.3.1 Overview

A detailed history of the agicultural development efforts of northern Cameroon
is presented in Chapter 5, including a more expansive description of the research system.
The following discussion presents a brief overview of both the Maroua research center,
and the research specific to cowpea and sorghum.

In 1974, the government "nationalized" the research system, creating the Oflice
National de la Recherche Scientifique et Technique (ONAREST) as a national umbrella
organization for agricultural research throughout the country. Since 1974, the

government has restructured its research system several times. Currently (1991),

 

17
agonomic research is conducted by the Institut de Recherche Agronomique (IRA) within
the Ministére de I ’Enseignement Superieur, de I Tnfonnatique et de la Recherche Scientifique
(MESIRES).

While the current agicultural research system is organized along major
ecological zones, with one research center per zone, budgeting and stafl'ing for these
centers are organized on a commodity basis. At the Maroua center, research units’
have been established to address production constraints for the principal cash and food

crops of northern Cameroon-cotton, sorghum, millet, rice, peanuts and mm.

 

2.3.2 Sorghum and Cowpea Research

The sorghum and mu units primarily screen varieties and test various
agonomic and post-harvest technologies. Sources of plant material for screening include
both promising local farmers’ varieties and exotic varieties. Introduced varieties are
distributed regionally for multi-locational evaluation by the International Institute for
Tropical Agiculture (IITA), the International Center for Research in Semi-Arid Tropics
(ICRISAT), the Semi-Arid Food Grain Research and Development project
(SAF GRAD), and the Bean/Cowpea Collaborative Support Project (CRSP). Although a
sorghum breeding progam was initiated in 1982, none of the developed hybrids were
released to farmers and the breeding progam was significantly scaled down after 1988.
W research initially focused on screening local and introduced cultivars. In 1987,

this research agenda shifted to identifying improved post-harvest storage technologies

 

7Maroua currently has research units for cotton breeding, cotton entomology,
sorghum and pearl millet breeding, sorghum and pearl millet agonomy, conea
agonomy, peanut breeding and agonomy, rice agonomy, farming systems research and
extension, and soil science (1991).

18

and to establishing a breeding progam to develop high-yielding cowPea varieties with
tolerance to the storage pest bruchids (Callasobruchus maculatus).
2.3.3 Staffing and Financing

Historically, a combination of expatriate and host country nationals have staffed
the research system. Initially, senior research staff were expatriates, employed by
I’Institut de Recherches Agronomiques Dopicales et des Cultures Vivrieres (IRAT), I’Institut
de Recherche sur le Coton et Fibres Textiles (IRCT), or donor projects while mid-level
staff and hourly workers were Cameroonian. Today, Cameroonians hold many of the
senior staff positions due, in part, to resources provided by USAID to train nationals in
the US. at the masters and doctorate level. Since the mid-19603, increased state and
donor funding and training opportunities have enabled the Maroua center to expand its

scientific staff (Table 2.4)’ and broaden its disciplinary mix. A

Table 2.4 Number of Senior Researchers, IRA Research
Center, Maroua, Cameroon, Selected Years.

 

 

 

 

 

 

 

 

 

 

 

 

Year Expatriate Cameroonian Total

1967 4 0 4

1977 6 3 9

1981 7 4 11

1987 15 15 30

1991 7 19 26
Source: Various IRAT and IRA documents.

 

'Ihe totals reported in Table 2.6 include researchers directly employed by IRA and
those affiliated with IRA through donor projects.

19
The research system has been funded jointly by the Cameroonian government

and donor projects. In recent years, the Cameroonian government has paid basic
operating costs (e.g., electricity, fuel, water), some capital improvements, and salaries for
Cameroonian researchers. Donor projects have usually financed equipment, vehicles,
capital improvements, staff training, and the salaries of expatriate staff.

Typically, donors have given priority to specific commodities. The French,
through the Caisse Centrale de Cooperation Economique (CCCE) and the Fonds d’Aide et
Cooperation (FAC) have supported most of the cotton research. SAF GRAD, the
European Economic Community (EEC) Development Fund, the World Bank, and
various national governments have funded food crop research. The United States, a
major supporter of food crop research, has provided (1979 to 1994) $46.7 million
through the National Cereals Research and Extension Project (NCRE), plus an

additional $1.97 million (1981-1992) through the Bean/Cowpea CRSP”.

2.4 Agicultural Extension

In northern Cameroon, extension-type activities are implemented by the farming
systems research team at IRA-Maroua, by MINAGRI’s extension service, and by the
internal extension system of the regional cotton parastatal, Ia Société de Développement
du Coton (SODECOTON).
2.4.1 IRA’s Farming System Research Efforts

Since 1979, the research system in northern Cameroon has embraced an on-farm

testing approach as a means of linking research to extension. The Center’s pro-extension

 

’Ihe NCRE progam supports research throughout Cameroon, whereas the CRSP
research is conducted only through the Maroua center.

20
progam was first organized under a SAFGRAD project (Joint Project No. 31). In 1986,

on-farm testing was institutionalized into the national agicultural research system when
the NCRE created the Testing and Liaison Units (TLU’s) at four of IRA’s research
stations, including the Maroua Center.”
2.4.2 MINAGRI’s Extension Efforts

The Ministry of Agiculture supports a structured extension progam, staffed by
Cameroonian civil servants. The various facets of extension, including farmer goups,
demonstration plots, credit progams, market outlets, and data gathering, have been
implemented with varying degees of success. Most key informants within the research
and extension complex reported that these efforts have had a negligible impact, due to
financial and logistic constraints within MINAGRI.u

In 1990, the World Bank funded the "National Agicultural Extension and
Training Project” to supplement MINAGRI’s extension progam throughout Cameroon.
In northern Cameroon, this project, based on the World Bank’s Training and Visitation
(T&V) model, targets the extension areas where SODECOTON is not present.
2.4.3 SODECOTON Extension Efforts

Since its creation forty years ago, SODECOTON has invested heavily in both
training farmers to use improved farming techniques and technologies, and training a
large corps of village level extension agents ('moniteurs'). In the mid-19803,
SODECOTON employed over 1,000 agents, although by 1990 this number had fallen to

 

10In the early 19803, a prototype TLU was established at the Bambui Research
Center in the Northwest Province, operating for several years before other TLUs were
created.

11Based on personal interviews with USAID contractors, IRA researchers and
SODECOTON personnel.

21
about 500 due to internal restructuring and financial constraints. Originally,

SODECOTON’s extension efforts targeted only cotton, but in the mid-19703 the
parastatal expanded its focus to include some food crops (primarily goundnut as an
additional source of vegetable oil). Beginning in 1982, with the creation of the World
Bank-funded Projet Centre-Nora (PCN), SODECOTON’s food crop extension activities
expanded rapidly.

The 1984 Agricultural Census data reported that 87,200 of the 268,500 farms in
the Far North Province planted cotton, indicating that thirty-two percent of the
Province’s farmers were exposed to SODECOTON’s extension efforts. Parastatal
reports indicate that the number of farmers gowing cotton rose from 88,315 in 1986/ 87
to 91,425 farmers in 1987 / 88. Since SODECOTON makes no effort to estimate the total
number of farmers (cotton plus non-cotton) in their extension zone, it is difi'icult to use
SODECOTON data to accurately estimate the percentage of farmers in this zone who
are exposed to SODECOTON’s extension system.

Data on the current number of farmers gowing cotton may underestimate
SODECOTON’s effect on agiculture since the number of farmers who once gew
cotton, but now do not, is unknown. Farmers may drop out of SODECOTON’S progam
for a variety of reasons, including SODECOTON’s dissatisfaction with the farmer’s
performance, the farmer’s dissatisfaction with the progam, and logistical constraints such
as inputs being unavailable when needed. At a minimum, more than one-third of all
farms in any given year have had some exposure to SODECOTON’s extension system-
either as a current or former cotton farmer or as a relative or neighbor of someone
gowing cotton. Most likely, over fifty percent of all farmers in the Far North Province

have been influenced by SODECOTUN’s extension system.

22
For farmers affiliated with SODECOTON, the extension progam has effectively

organized farmers into goups and has provided access to short-term credit, advice from
technicians, market outlets, and equipment, seed and chemicals.

Part of the success of SODECOTON’s extension system lies in its close linkages
to IRA, which were strengthened considerably by the PCN. As IRA generated
appropriate cotton and food crop technology, SODECOTON was readily available to
extend them through its extension network, provide feedback on their performance, and

suggest topics for future research. These relationships are discussed in detail in

Chapter 5.

CHAPTER 3 LITERATURE REVIEW AND METHODOLOGY

This chapter is divided into four sections. The first section reviews the
technology assessment literature. The second discusses the benefit-cost approach to
impact assessment, focusing on its application to cowpea and sorghum research in
northern Cameroon. The third describes the methods used in this study to analyze the
role and impact of institutions supporting cowpea and sorghum research. The fourth

summarizes how data were collected for this study.

3.1 Review of Literature
3.1.1 Introduction

Since the 19503, agicultural economists have sought to assess the impact of
agricultural research. T. W. Schultz (1953) and Z. Griliches (1958) report two of the
earliest efforts to determine the economic returns to public investments in agicultural
research. As Griliches noted, "expenditures on ‘research and development’ have gown
very rapidly in the last decade. Quantitatively, however, we know very little about the
results of these investments. We have some idea of how much we have spent but very
little of what we got in return (ibid, p. 419)." Thirty-five years later, the questions
related to assessing research investments are more focused and the methodologies
employed to estimate economic returns are more refined. Yet, the ability to quantify the
benefits of agicultural research is still limited by fundamental assumptions that, at times,
are little more than ”educated guesses”.

Anderson and Herdt noted two broad perspectives in assessing the impact of a

given research activity. "The first is more concerned with the mechanisms and process of

23

24
research. Observers with this perspective are interested primarily in what the direct

products of research have been (1990, p. 36).” In this case, the actual technologies (eg,
new varieties, new ago-chemicals) that are developed are considered the "impact" of
research. The second category is the more interesting, but paradoxically the more
complex issue of how those technologies affect the clientele of research. Issues relevant
to this line of inquiry include how a technology changed what is done on the farm, and
its impact on household nutrition and rural income. To address this second level of
inquiry, more comprehensive analytic approaches have been developed.
3.1.2 Methodologies

Researchers employ a wide range of approaches to assess impact. At one end of
the spectrum are those who limit their analysis to qualitative check lists (J ahnke, 1987;
J aeger, 1987; Busch, 1989). These approaches attempt to describe how products of a
research and / or extension system have influenced macro-economic and social indicators.
This methodology uses ”informed" descriptives to assess project implementation and may
attempt to identify research outputs. Evaluation is limited to speculative linkages of the
relationship between research outputs and changes at. the aggegate level (eg, changes in
GDP, employment rates, household incomes and diet, infant mortalities). At the other
end of the spectrum are those who employ detailed, quantitative econometric models to
calculate elasticities, supply and production functions, and the marginal rates of return
on investments (Griliches, 1964; Evenson et al., 1979; Lu et al., 1978).

Within this qualitative-to-quantitative continuum, there are a variety of methods
for describing and quantifying research and extension impact. For example, Evenson
specifies six "evaluation types", differentiated by whether or not economic values are

assigned to research outputs, and the type of statistical methodology applied in the

25
analysis. His six evaluation types are managerial, technical, invention studies (non-
statistical), invention studies (statistical), meta-studies (statistical), and studies on
economic consequence (1990, p.4).

Most published impact assessments attempt to quantify benefits as economic or
financial returns to “investments" in agicultural research. Echeverria (1990) goups
these methods into two general classes, the economic surplus approach (consumer-
producer surplus, benefit-cost, and index number methods) and the econometric
approach (production, profit, and supply functions and their derivations). The former
"estimates returns on investment (generally an average rate of return) by measuring the
change in consumer and producer surplus from a shift to the right in the supply curve
due to technological change (ibid, p. 2)."l2 The latter “treats research as a variable and
allows a marginal rate of return on investment to be calculated (ibid, p. 2)."13

Economists employ several different measures of profitability when evaluating an
investment’s return. The internal rate of return (IRR) is one such measure, as defined

below:

 

12 Examples include Akino and Hayami’s (1975) estimate of the returns to rice
research in Japan, and Enamul Haque et al.’s (1989) estimate on the returns to broiler
chicken research.

”Examples include Thirtle and Bottomley’s (1988) study of public investments in
research and Nagy’s (1978) assessment of rapeseed research in Canada.

26

 

Internal ROR The interest rate that equates the net present value of cash
flows to zero, as calculated by the equation:
' B: ' Ct .. o (3.1)
is! (I + r)‘

where 'B" and 'C' are the values of the benefit and cost streams
and "r' is the IRR, the interest rate that solves the equality.
There are "t' a 1 to "n" time periods in the analysis.
An IRR of zero indicates a return sufficient to cover the initial investment, but no more.
The IRR must be equal to or geater than the target rate of return (the opportunity cost
of capital) in order for the investment to be considered “profitable”.

Economists make further distinctions when measuring profitability, two pairs of
which are listed below. The first pair accounts for the difference between real and
nominal valuation. The second addresses the question of which investments are being
analyzed—total or marginal.

1. Financial ROR An IRR where the benefits and the costs are in terms of market
prices and domestic currency, unadjusted for shadow exchange
rates or inflation.

Economic ROR An IRR where the benefits and costs are valued in terms of
their opportunity costs, reflecting such factors as scarcity values

and shadow exchange rates.

2. Average ROR An ROR where the returns to the sum of total project
expenditures are calculated.

Marginal ROR An ROR where only the rate of return to the last, "marginal”
dollar invested is calculated.
Early impact studies focused on ex post assessments. In recent years, economists
have extended the ex post methods to develop ex ante methodologies to compare the
returns to alternative research strategies. For example, Coth and Bottomley (1989)

include a benefit-cost analysis as one of several methods proposed for setting priorities

27
in agicultural research. Similarly, the World Bank (Gittinger, 1982) regularly calculates

an expected rate of return for proposed projects. Applications of ex ante analysis to
specific research projects include dc Frahan’s (1990) estimates for farming systems
research in Mali, and Scré and Jarvis’ (1990) analysis of expected returns to pasture
research in Latin America.

Several economists have recently refined impact assessment methodology to
incorporate effects previously ignored. Evenson (1989, p 447) incorporates "spillin and
spillout" effects into his analysis. Prady and Wood (1991, p. 10) modify Evenson’s
approach by first differentiating research progams along ”their technology foci" and then
look at their spillover effects. Horton (1990, p. 52) identifies two types of impact,
production and institutional, noting that most of the literature to date has ignored the
institutional factors. Capalbo and Antle (1989, p. 458) incorporated social costs (cg,
pollution, soil erosion) in the calculation of the returns to research. While these
advances represent empirical improvements, they make the analysis increasingly more
complex and more dependent on the availability of extensive amounts of data.

3.1.3 An Underlying Assumption

All of these methods assume the existence of a reliable data base. Generally,
there is a positive correlation between data detail and quality, and the degee of
methodological sophistication. The more statistically advanced the model, the geatcr
the data needs (and potentially the geatcr the cost to collect these data). Out of
concern about the costs of ROR studies, Anderson and Herdt (1990, p. 42) have tried to
estimate the impact of an impact assessment study. Although their results were
inconclusive, their efforts reflect the need to justify impact assessment, given its potential

high costs. On the other hand, Little and Mirrlees (1990) quite strongly advocate the

28
need for and benefits from benefit-cost analysis. Furthermore, there are theoretical

tradeofls between alternative methods for assessing impact. Impact studies which
quantify and value the returns to research generate either an average or a marginal rate
of return. With an average rate of return, the "benefits can only be calculated for the
total investment, rather than small increases in the research budget. Thus, policy makers
can only use the ARR technique to estimate the return to an entire project expenditure-
not how much to increase research spending (Oehmke, et al., p. 6)“, or how to allocate
among projects. In theory, studies that generate a marginal rate of return help
determine the optimal allocation of funds for the research project. However, marginal
RORs require highly accurate and detailed time-series data on inputs, outputs and
research expenditures.

For a project that is already completed or is going through a final evaluation
phase, an ARR approach is generally regarded as appropriate. On the other hand,
managers of on-going research activities who regularly allocate capital (financial and
human) will generally find the MRR approach more informative. Ultimately, for the
practitioner, data availability usually dictates the methodology employed when assessing

the impact of a given project.

3.2 Benefit-Cost Approach
3.2.1 Overview

The results of benefit-cost analysis are reported as benefit-cost ratios, net present
values, or internal rates of return. All of these measures use the same underlying data
base, comprised of two discounted time series-the benefit stream and the cost stream.

The benefit (cost) stream is a time series of benefits (costs), valued in some currency.

29
Because of the time-value of money, these streams are discounted, usually based on

assumptions about the opportunity costs of capital.
3.2.2 Advantages of Benefit-Cost Analysis

Choosing a method for assessing impact depends on several key factors. These
include the availability of data, the objectives of the assessment exercise, and the timing
of the study.

Data availability in northern Cameroon is discussed in detail in the fourth
section of this chapter. Available data were sufficient for some but not all forms of
quantitative analysis. Data could be organized into benefit and cost streams, but data
were not sufficient to accurately estimate demand and supply curves (and elasticities) for
the cowpea and sorghum subsectors. Without estimates of elasticities, some types of
econometric analysis were impossible, thus eliminating some types of quantitative
analyses from consideration (cg, index number approach).

All relevant expenditures for this study have occurred in the past. The analytic
objective is limited to estimating the average rate of return over the sum of these
historical expenditures. Since the marginal rate of return to incremental increases in
past or present expenditures was not an objective of the study“, benefit-cost analysis,
which generates an ARR, satisfies this study’s needs.

Thus, this study calculates benefit and cost streams for W and sorghum
research and extension in northern Cameroon, which are used to calculate an internal

rate of return for the historical investments made in research and extension (Chapter 4).

 

“Karanja (1992) provides an example of calculating an MRR using historic data on
maize research in Kenya.

29

Because of the time—value of money, these streams are dhcountcd, usually based on
assumptions about the opportunity costs of capital.
3.2.2 Advantages of Benefit-Cost Analysis

Choosing a method for assessing impact depends on several key factors. These
include the availability of data, the objectives of the assessment exercise, and the timing
of the study.

Data availability in northern Cameroon is discussed in detail in the fourth
section of this chapter. Available data were sufficient for some but not all forms of
quantitative analysis. Data could be organized into benefit and cost streams, but data
were not sufficient to accurately estimate demand and supply curves (and elasticities) for
the W and sorghum subsectors. Without estimates of elasticities, some types of
econometric analysis were impossible, thus eliminating some types of quantitative
analyses from consideration (cg, index number approach).

All relevant expenditures for this study have occurred in the past. The analytic
objective is limited to estimating the average rate of return over the sum of these
historical expenditures. Since the marginal rate of return to incremental increases in
past or present expenditures was not an objective of the study“, benefit-cost analysis,
which generates an ARR, satisfies this study’s needs.

Thus, this study calculates benefit and cost streams for W and sorghum
research and extension in northern Cameroon, which are used to calculate an internal

rate of return for the historical investments made in research and extension (Chapter 4).

 

l‘Karanja (1992) provides an example of calculating an MRR using historic data on
maize research in Kenya.

30
3.2.3 Potential Disadvantages of Benefit-Cost Analysis

Two factors potentially limit the validity and reliability of the benefit-cost
approach for assessing impact. First, it uses several simplifying assumptions about prices,
slopes of the demand and supply curves, and the extension of benefits into the future.
Second, since the results are usually aggegatcd, distributional issues are typically not
evaluated.

With the benefit-cost approach, prices are usually held constant when quantifying
the benefits of an improved technology. This suggests that supply is perfectly inelastic
and demand is perfectly elastic. This assumption implies that supply can be increased
without lowering prices or flooding the market with overproduction. For relatively small
increases in total supply, as was the case for the technologies extended in northern A
Cameroon, this assumption is reasonable.

The second set of assumptions concerns the extension of the benefit stream into
the future. Although discounting effectively negates benefits in the distant future
(geatcr than 20 years), benefits in the immediate future must be projected. Thus,
assumptions regarding the lifetime of the technology and the degee of adoption (the
adoption ceiling) are required. As Evenson notes, ”for most imputation-accounting
studies the analyst must make some assumptions regarding the continuation of the
benefit stream beyond the period of analysis (1991, p. 12).” The alternative is to assume
all benefits stop as of today, which clearly underestimates total benefits. Hence, well-

informed assumptions about the short-term future are the preferred alternative".

 

I’Extending benefit streams into the future is not unique to benefit-cost analysis.
This is an issue for nearly all assessment methods.

31
Evenson (1991) notes an additional disadvantage to the benefit-cost approach.

In general, both consumers (due to decreased prices) and producers (due to increased
yields) benefit from a yield-enhancing technology. The benefit-cost approach determines
total net gains, but it generally does not determine how these gains are distributed
among the various segments of society. Hence, in evaluating the gains from W and
sorghum research, qualitative statements about their distribution are subjective. This
limits critical aspects of impact assessment since tradeoffs between such groups as rural
producers and urban consumers, and men and women in households are not fully
identified. To partially address this problem, this study combines the benefit-cost
approach with an institutional analysis that employs qualitative evaluation techniques to
assess how improved technologies affect different segments of northern Cameroon -

society.

3.3 An Approach to Institutional Analysis
3.3.1 Overview

Schmid broadly defines "institutions” as ”sets of ordered relationships among
people that define their rights, their exposure to the rights of others, their privileges, and
their responsibilities (1987, p. 6).” In Cameroon, important institutions that affected the
productivity of research included the government’s system of research and extension (i.e.,
IRA, MINAGRI), input suppliers like the NCSM Project and SODECOTON, output
markets, donor projects, and the government’s policies towards food crop marketing (de
facto Iaissez—faire).

In the context of impact assessment, institutional analysis examines how

institutions affect the benefit and cost streams. Institutional analysis is generally done by

32
comparing a system’s performance with and then without a given institution. A specific
case of this is 'before/after'I analysis which compares a system’s performance before a
given set of institutional arrangements were created to the system’s performance once
this set is operative.
3.3.2 The Contribution of Institutional Analysis

Institutional analysis can help identify factors that contributed to the productivity
and "success” of a new technology. Quantitative analyses (eg., IRR calculations) simply
estimate the financial and/ or economic returns to investments. Policy decisions based
solely on quantitative results are limited to choices between alternative investments with
high, low, or negative returns. Qualitative analyses (cg, institutional analyses) seek to
explain why an investment had high, low, or negative returns. With these insights, the
policy choice set is greatly expanded to include policies that alter the potential returns of
investments. For example, an investment which historically has had low returns still may
be investment-worthy, if institutional constraints that caused the low returns are altered
by policy changes.

Qualitative analyses may also help to explain how returns are distributed. For
example, investments with high returns that benefit only a small group may be valued
differently from investments with high returns that benefit a much broader constituency.
Hence, analysis to identify the beneficiaries of the research and extension system of
northern Cameroon is an important complement to calculating the net benefits of the
system.

This study first identifies key institutions of northern Cameroon and describes
their histories. The evolution of the research and extension system is then synthesized,

in part, with a detailed time-line of the development of the key institutions. Qualitative

32
comparing a system’s performance with and then without a given institution. A specific
case of this is 'before/after' analysis which compares a system’s performance before a
given set of institutional arrangements were created to the system’s performance once
this set is operative.
, 3.3.2 The Contribution of Institutional Analysis

Institutional analysis can help identify factors that contributed to the productivity
and ”success“ of a new technology. Quantitative analyses (eg., IRR calculations) simply
estimate the financial and/ or economic returns to investments. Policy decisions based
solely on quantitative results are limited to choices between alternative investments with
high, low, or negative returns. Qualitative analyses (eg., institutional analyses) seek to
explain why an investment had high, low, or negative returns. With these insights, the
policy choice set is greatly expanded to include policies that alter the potential returns of
investments. For example, an investment which historically has had low returns still may
be investment-worthy, if institutional constraints that caused the low returns are altered
by Policy chainsa-

Qualitative analyses may also help to explain how returns are distributed. For
example, investments with high returns that benefit only a small group may be valued
difierendy from investments with high returns that benefit a much broader constituency.
Hence, analysis to identify the beneficiaries of the research and extension system of
northern Cameroon is an important complement to calculating the net benefits of the
system.

This study first identifies key institutions of northern Cameroon and describes
their histories. The evolution of the research and extension system is then synthesized,

in part, with a detailed time-line of the development of the key institutions. Qualitative

33
descriptive data on the system are based on interviews with key informants and syntheses

of project documents. The focus of the analysis is to evaluate the linkages between each
component of the system and how these linkages affected the system’s performance.
This analysis is detailed in Chapter 5.

3.4 Data Collection

Impact assessment literature seldom discusses data constraints, which is a critical
failing since available data usually determine the assessment methodology employed.
The following details this study’s data collection process, including a description of the
available data, how they were compiled and their reliability.
3.4.1 Data Availability

This study used the ”rapid appraisal” methodoloy recommended by Holtzman
(1986). The approach, which relies heavily on existing documents and oral histories
provided by system participants, was used since primary data collection, based on long-
term monitoring of the system and formal surveying of sector participants, was beyond
the scope of this study. Data for the research and extension system of northern
Cameroon were obtained primarily from project reports (usually annual) and research
summaries. When these sources failed to provide sufficient detail for the needs of the
analysis, "key informants“ were interviewed.

Documents detailing research activities and resources employed were available
from the research libraries of the IRA-Maroua research center, EEC-Garoua’s and

USAID-Yaoundé’s reference rooms, and from the personal libraries of researchers and

 

“Key informants were individuals within the research and extension system and the
W and sorghum subsectors (eg., plant breeders, extension agents, farmers, grain
merchants) who were knowledgeable about the data in question.

34
administrators of SODECOTON, MINAGRI, IRA, and the NCRE project. The data

needs spanned thirteen years of research and extension efforts (1979 to 1991) that were
associated with the work of five large development projects". Considerable effort was
required to find copies of all relevant documents, since some were over ten years old.
Also, since annual reports and research summaries constituted the bulk of available
documents, interviews with key informants were necessary to meet information gaps.

In an attempt to identify all available data, the interview process, when possible,
was iterative. During the initial interview, key informants in the research and extension
system were provided an overview of the study and data needs, and documents from the
interviewee’s personal library were solicited. After reviewing these documents, a follow-
up interview was conducted as the documents were being returned.

Plant breeders and agronomists provided information about the local farming
systems and technologies that had been developed and extended. Administrators
provided much of the institutional history, as well as data on research costs. Extension
and input delivery personnel provided information on costs, rates of adoption, and
institutional linkages within and outside the SODECOTON system. Farmers, grain
merchants, and village extension workers were interviewed to confirm the opinions of
research and extension personnel.

3.4.2 Data Base Compilation

As stated above, sufficient data were available for calculating the benefit and cost

streams necessary for ROR analysis. However, the available data had not been collected

and organized specifically for this type of study. Thus, this ”secondary data” had to be

 

1"The projects were SAFGRAD’s Joint Project No. 31, the World Bank’s Projet
Centre-Nerd, and USAID’s NCRE project, Bean /Cowpea CRSP, and NCSM Project.

35
sorted and compiled into data sets appropriate for analysis. These sets included time

series data on: (1) general economic parameters such as the inflation rate, exchange
rates, and interest rates, (2) the benefits of the research-extension system, and (3) the
costs of creating and maintaining this system.

For example, estimating the benefit streams required a broad range of data,
including information on land in production, prices, adoption rates, and yield gains and
changes in input mixes resulting from the adoption of extended technologies. These data
were compiled from project reports, Agricultural Census reports, and interviews with key
informants. Similarly, estimating the cost streams required detailed time series of
expenditures. Since these series were not always available, proxies and estimates of costs
had to be compiled from project documents and key informant interviews. The entire
process is detailed in Chapter 4.

3.4.3 Data Reliability

Although information gathered during interviews with key informants is critical
to this study, the analysis relies heavily on secondary data. For example, the benefit
streams include estimates of land in production of various crops, which are based on
census estimates, extrapolated from a sample of farmers’ fields judged to be
”representative“ of all farms in the Far North Province. Sampling and aggregation errors
for this methodology are likely, given the limited resources for survey work in northern
Cameroon. Other components of the benefit and cost streams that represent ”best
estimates" include the commodity and input price data, the adoption rates, and extension
costs. Thus, given the heavy reliance on secondary data, the integrity of this study’s
results depends, in part, on the reliability of this secondary data. The specific limitations
of the data sets and the assumptions made to correct for their limitations are included in

Chapter 4.

CHAPTER 4 RATE OF RETURN ANALYSIS

This chapter is divided into two major sections, one for each of the commodities
studied. Each section (1) specifies the cost and benefit streams, (2) describes the
components which comprise each stream, (3) presents the underlying data, their sources
and their limitations, (4) specifies the assumptions made in interpreting these data, and
finally, (5) reviews critical parameters of the analysis, applying sensitivity analysis to

examine the effects of modifying their estimated values.

4.1 General Approach

The cost streams represent estimates of annual research and extension
expenditures by donor projects and host country programs. The benefit streams are
estimates of the annual dollar value of project benefits-calculated as the market value of
gross benefits, minus the value of additional on-farm costs of using the technologies.
Gross benefits are calculated as the product of gains in yield from the improved
technologies, adoption rates, and land in production.

Data for the benefit-cost analysis are presented in nominal US dollars, having
been converted from fcfa when necessary”. Over the entire period of the analysis, the
exchange rate between the fcfa and the French franc was fixed at 50 fcfa per one French
franc. Thus, fluctuations in the value of the fcfa should simply reflected changes in the '
exchange rate between the US dollar and French franc. Some research (Salinger and

Stryker, 1991) indicates that the fcfa is overvalued. While no effort was made to

 

”Prady and Rosenboon note that ”there is...no option but to convert research
expenditures measured in current local currency units into some numéraire currency or
unit of measurement (1989, p.24)"

36

37
calculate a shadow exchange rate for the base runs, the exchange rate was included as a

variable in the sensitivity analysis.

A second simplifying assumption about costs and benefits was made regarding
the inflation rate applicable to the subsistence economy of northern Cameroon. While
inflation rates (IMF and World Bank internal documents) have been calculated for the
urban centers of southern Cameroon (Yaounde and Douala), these data have little
applicability to the economy of subsistence farming 800 kilometers to the north. Further,
available market prices (MINAGRI Annual Reports) indicate no discernable price
trends. Also, since weather effects dominate price fluctuations in northern Cameroon, it
is impossible to identify short-term inflationary trends in price. Hence, the base runs of

the analyses are estimates without any adjustment for inflation.

4.2 Cowpea
4.2.1 Overview of Improved Cowpea Technologies

Cowpea, a crop indigenous to northern Cameroon, is traditionally intercropped
with either sorghum or millet. Both the leaves and grain pods are harvested for food.
Although cowpea leaf consumption fills an important niche in the region’s household
food security, cowpea research (from 1979 to 1987) focused almost exclusively on
increasing grain yields since many farmers already grew local varieties that produced
high leaf yields.

The first technology package developed by the research system included a new
cowpea variety, M 3236 061. This indeterminant, medium cycle (75 to 80 days to
maturity) cultivar was selected from IITA regional screening trials for its high yield

potential, white grain color, and insect (thrips) tolerance. IRA and SODECOTON

38

recommended that farmers monocrop the variety on a quarter-hectare plot and, when
possible, treat the standing crop with insecticide.

Although TVX 3236 was first 'extended' to farmers in 1980 through
SAFGRAD’s on-farm testing program, SODECOTON did not begin to extend the
variety widely until 1984. Widespread extension was facilitated by the North Cameroon
Seed Multiplication Project, which produced and sold approximately twenty metric tons
of TVX 3236 from 1984 to 1986. SODECOTON continued to recommend and extend
the 'I'VX package” through the 1987 growing season. In addition, IRA introduced Ife
Brown (a local Nigerian cultivar) and WA (a local Cameroonian cultivar from the
Moutourwa area) in 1985 and 1986-1987, respectively, two promising cultivars identified
through SAF GRAD / CRSP screening trials.

During this period (1980-1986), researchers and extension workers observed
significant (sometimes total) storage losses due to bruchid infestations. As a result,
SODECOTON reduced the recommended plot size from a quarter to an eighth of an
hectare. SODECOTON’s contention was that until the storage constraints could be
lessened, farmers should grow COWpea as a garden /compound food crop for the hungry
season, not as a commercial grain crop.

In 1987, IRA released two sister lines from IIT‘A with several advantages over
TVX 3236 including comparable yield, larger grain size, significantly less shattering of
seed pods, and most important, greater tolerance to bruchids. These two varieties, BR]
and BR2 (IITA cultivars IT81D-985 and IT81D-994, respectively), were judged
sufliciently tolerant to bruchids to allow farmers to store cowpea grains for an additional
month before bruchid damage became significant. Table 4.1 summarizes the chronology

of extension recommendations.

39

Table 4.1 Extension Recommendations for Cowpea, Northern Cameroon, 1984
to 1991.

 

 

 

 

 

 

 

 

 

 

Improved Technologies 1988 to 1991
Monocropped W use see use use use
Insecticide ‘*" *"3 ttt no use
1/4 ha 'block' plot see ease ems use use
1/8 ha ”garden” plot mu ems
Variety TVX 3236 *" .tt else
Variety Ife Brown u:

Variety VYA ease ems as”
Varieties BRl and BR2 at an

 

 

 

 

 

 

 

 

Source: Annual reports for IRA, CRSP, SODECOTON, and SAF GRAD.

""" indicates the years in which the improved technology was extended to farmers.

Since 1987, researchers have continued to advise farmers to plant W as a
monocrop in quarter-hectare plots, sowing BRl and BR2, and applying 2 to 3 insecticide
sprayings. In addition, in recognition of the importance of post-harvest losses, the
research agenda shifted to give geater priority to developing improved gain storage
technologies and to establishing a breeding progam directed, in part, at increasing
tolerance to storage pests (bruchids). However, as this research initiative is beyond the
scope of this study, its costs and impacts are not included in the benefit-cost streams.
4.2.2 Cowpea Cost Stream
E . . l C E

The package extended to farmers consisted of an improved variety (first

TVX 3236, then BR] and BR2) and a set protocol that differed considerably from

40
traditional farming practices. Three donor projects (the Bean /Cowpea CRSP,
SAFGRAD’s LR No. 31, and the NCRE project) and two host country institutions (IRA
and SODECOTON) contributed to the development of this technology package. Annual
expenditures attributable to W research and extension are compiled and detailed
below for each of the five institutions involved.

Since research specific to the improved technoloy was initiated in 1979, moved
into on-farm testing as a technological package in 1981 (for TVX 3236, in 1984 for BRl
and BR2) and extended to farmers in 1984 (for TVX 3236, in 1987 for BRl and BR2),
only costs incurred during this nine-year period are included in the cost stream.
W

SAFGRAD Joint Project No. 31, which began in 1979, was mandated to conduct
both on-station and on-farm research on four food crops widely gown in the semi-arid
environment: sorghum, millet, maize, and cowpea. The project employed one expatriate
researcher during each of the nine years of its existence. While IRA provided support
staff and some research funding, all other operating costs were covered by SAP GRAD.

SAFGRAD project costs for the period 1979 to 1983 are summarized in a final
report submitted by Owen Gwathmey, the first of two expatriate researchers assigned to
the project. The report provides five-year totals in nominal dollars by main expenditure
categories.

Since cowpea was only one of four crops researched by the project, it was
necessary to adjust total project expenditures to reflect the commodity’s share of the
total project’s cost stream. Based on interviews with both Gwathmey and Martin
F obasso, Gwathmey’s Cameroonian counterpart, the cowpea share of total expenditures

was estimated at 25%. The adjusted five-year totals were converted into annual

41

expenditures by using the following assumptions: for salary allowances, housing, and
other direct costs, an annual 4% "institutionalized" increase was assumed; for travel, per
diem, and shipping and storage, 35% shares of the total were allocated to the first and
last years, and 10% shares for the three middle years; for equipment, supplies and
materials, a descending share series was assumed with 35% of the total allocated to the
first year, followed by annual shares of 30, 15, 15 and 5% to the subsequent years,
respectively. The resulting cost stream for the period 1979-83 is reported Table 4.2.

Jerry Johnson, the second expatriate to work on the project, provided estimates
of project and salary costs (nominal US dollars) for the years 1984 to 1987 (personal
interview). These data are drawn from his personal files, directly from past contracts
and canceled voucher reports. Since these data were not available by detailed .
expenditure category, they are reported in Table 4.2 as salary allowances and non-salary
expenditures. Personal services contracts varied from 13 to 27 months and never
coincided with the calendar year. Hence, to estimate annual contract costs, the
expenditures associated with a given contract are divided by the number of months for
which the contract was written and then summed over the calendar year. The W
share of total expenditures was estimated at 25%, based on interviews with Johnson.
The resulting cost stream for the 1984-87 period is also reported in Table 4.2.
W

The CRSP began work in northern Cameroon in 1982, in affiliation with two
host institutions (IRA and SODECOTON) and the on-going SAF GRAD project. Cost
data for CRSP activities were provided by the Bean / Cowpea CRSP management office

at Michigan State University. These annual expenditures were listed by expenditure

42
category in nominal dollars. Since all direct fiscal outlays were ultimately authorized and

paid for through this office, these cost data are considered complete (Table 4.3).

One caveat is that from 1982 to 1986 the project contractor (the University of
Georgia) apparently paid the salary and benefits of the project’s principal investigator,
but these costs are not included in the contractor’s expenditure totals reported in the
original CRSP documents. Similarly, in 1987, the local USAID mission in Cameroon,
instead of the contractor, paid this expense. Hence, these costs had to be estimated
from available data and added to the CRSP cost totals. Information on the salary and
benefits of the principal investigator was only available for 1989". By extrapolating
backwards and assuming a 5% per annum increase in salary costs, the PPS salary history
was estimated for the period 1982 to 1987. '

A second caveat is that "US Matched Funds”, as reported by the CRSP
management office, are not included in the CRSP cost stream used in this analysis.
These funds are excluded because they are not part of the CRSP budget, rather they
mainly represent the value of services provided to the CRSP by the US institutions.
Although the research efforts in northern Cameroon were indirectly supported by these
financial outlays, most of the resulting benefits were not realized in Cameroon. Since
these benefits are not included in the benefit stream of the analysis, their costs are also

excluded.

 

1’Bean/COWpea CRSP - Purdue University (internal document provided by Katy
Ibrahim) reported that the 1989 salary and benefits of the principal investigator, Dr.
Mofi Ta’Ama, totaled $89,544. Annual costs in previous years were not available,
although this expenditure was reported to be relatively constant over the lifetime of the
project, increasing only slightly (3 to 5%) each year.

8.365%» $62

:o .858» 38 cuckoo 05 5 v80: as .880me mews—one can 22— mmoz can? 53 5 .028 u. m 2: 5a.— 8 vogue
22$. .soea mmoz 2: 53 .58.. s. a .82 5 Eu 9 838‘s 32E... 3.. sees 22% 2:.
“Convene shame—5&0 053% .3 $333.5 8:. See 855?. as.

A8353 .2592.» £35323 ”:3: no .80 $3 8% be?“ 05 E tuna—2: 2a amoeba.

£852. .n .23 .Smanom .2 $055550 .0 .53 303.85 maeofiauon 603$ ”825m

 

 

 

 

 

43

 

 

 

 

 

 

 

movfn «new»; omn.~5 cwm.5_ go «3.2 :3 wmwd .2505
v5.3 v8.3 so a: a: a: a: a: $3
wmcfim 53.3 :56 a: a: a: a: a: £2
8an Shea 93: a: a: a: a: a: 3%
gain 85.3 52.3 a: a: a: a: an: $3
ncné Sud v8.5 new on“ :m.m vow SYN mae—
535 ~5w.w 356 Sea 3; main 3: ewe 33
53.9 own.» awed 5oc.~ e2 39m 3_ one 33
3&5" mm; 52.5 Ema m2 «:3 8“ one 83
25.3 8&5 SQS «.86 mm— nmwfi vow Stu e53
momeonxm— mites: 380
3.9.00 8530—2 ba_~m.:oz mom—e96 “8:9 ems—Sm on. E05 com
Eek lbs—am .8825 EoE&=vm e050 wfimsom meant—w a. $22.5 30>

 

 

 

 

 

 

 

 

 

.53 2 53 £022.80 6.5.32
.8 62 60.3... 32 92053 geese 8980 .63 :5...on 832%...“— EB. Begum a... use.

 

5503:: annum 53:02 .0050 E25952 ammo Sea 83883 3285."— "02:8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

53.83 53.59 3N.S amino" www.mfl 32.3. :«a: £805
«9.63 398 3a.; 35.53 58 326m wvm.5 5w3
mam.~5~ 813 o 3.8.3. can.“ ”.363 33.5 83
nmm.weu ch.8_ 9 83¢ 3o.§ «and... 3 flan B3
n8.~mm 15.3 c not? 8er $5.83 5%.; $3
2668 85.3 a mat; mince main v8.3 83
333 m5~.~5 a $0.3 c 35.9. gew $3

32130 8:50am
QZmD a. 380 “3.082
380 =o_8_2 280 ~0th eczema :55 com
38.5 659.com 180.— 825: .050 EoEchm Q .385 “50>
.53 2 $3 .595an .3982 J3me...—
eoasm geese oéezu=oo «eggs. .63 1.5.85 BEBE Bop 33.53 3 03.9

 

45
W

The NCRE project began in 1982, but had a very limited role in the development
and extension of cowpea technologies during the early 1980s. Given that the project was
mandated to research cereals and that both the CRSP and SAF GRAD project were
researching cowpea, there was little need for the involvement of the NCRE staff and
resources beyond informal collaboration with on-farm tests. However, in 1986, as the
SAFGRAD project was drawing to a close, and as the NCRE project was expanding into
on-farm testing, the NCRE project negotiated a technical ”buy-in" (during the last two
years of the SAP GRAD project) which financed some on-farm research costs. The share
of these operating expenses associated with cowpea research (estimated at 25% of total
costs incurred) was approximately $4,890 in 1986 and $9,780 in 1987. '
W

IRA, as the host country research institution, assumed part of the cost of
developing the new cowpea technologies-including the salaries of host country research
staff and unskilled labor, and some operating expenses (eg., fuel, electricity, water, office
materials, per diem, temporary hires).

For years 1982 to 1987, the CRSP management office reported estimates of
financial contributions (in nominal $US) to the project by "host country institutions”.
Assuming that these estimates represent total outlays by IRA for the cowpea project, the
data (Table 4.4) are added directly to the cost stream.

Prior to 1982, IRA collaborated with the SAF GRAD project, although no
document quantifying the value of this contribution was found. To make up for this
deficiency, an annual IRA cost stream was created, based on IRA documents reporting

staff and operating costs. A history of IRA staff (researchers and administrators)

46

involved in on-farm testing and cowpea research and their salaries was constructed from
IRA documents. The history was then converted to an annual cost stream of IRA salary
costs. Combining data provided by IRA’s accountant at the Maroua research station
with data from various documents, operating expenses were also estimated. The
resulting cost stream (Table 4.4) underestimates IRA’s contribution since it does not
include 'in-kind‘ costs such as land and buildings, although these costs are included in
the CRSP estimates.

For the benefit-cost base run, the CRSP cost estimates were used during the
years for which they were available (1982-1987). For earlier years (1979-1981), the
constructed cost stream was used.

W

As part of its general activities, SODECOTON maintains a large extension
network. The adoption of the cowpea package and its subsequent impact is, in part,
dependent on the extension and distribution system of SODECOTON.

SODECOTON ofl'icials estimated that their extension and management staff
spent approximately 10% of their time on food crops. Since agricultural census data
indicate that cowpea is planted to 5% of all food crop land in the Far North Province, it
was assumed that 5% of food crop extension costs were allocated to the extension of
cowpea technologies. Further, since available data were only for staff salaries and
benefits, an overhead adjustment (20% of the costs of salaries and staff benefits) was
included in the cost stream to account for such costs as fuel, supplies, equipment, and
materials. Hence, the cost of W extension activities was estimated as total staff

costs, Mo multiplied by the factor CCE (Equation 4.1).

47

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.4 Estimated Total Expenditures in Nominal SUS for the IRA-
Cowpea Program, Maroua, Cameroon, 1982-1987, Costs
Estimates from Two Sources.
Year Stafl' & Operating Costs Total Costs Estimated by
Estimated from IRA Bean / Cowpea CRSP
documents
1979 10,771 na'
1980 11,001 na
1981 13,278 na
1982 20,687 33,478
1983 21,022 41,847
1984 32,451 58,585
1985 35,493 55,103
1986 46,116 26,462
1987 45,954 84,954
Source: Financial Summaries from CRSP Management Office, Michigan

State University and from IRA documents and interviews with

the IRA-Maroua accountant.

‘na signifies that data are "not available / estimated” since the Bean / Cowpea
CRSP did not exist in northern Cameroon prior to 1982.

 

48

ch‘M:*(Src*Acr*Sov)

= M, #1 Ca (41)
= M, t 0.006

where:

TC. = Total extension costs for mpea, year t

M, - SODECOTON agricultural staff costs, year t

can = CUWpea extension cost factor - 0.006

SK. 8 Percent of extension staff costs allocated to food crops - 0.10

AC, 8 Percent of food crop extension costs alloeated to cowpea = 0.05

Sov 8 Overhead adjustment 8 1.20

Since the annual cost figures were reported in nominal fcfa, they were converted
to nominal SUS by using annual average market exchange rates reported by the IMF.
Extension costs were estimated for seven years (1981 to 1987), given that the SAFGRAD
project and the CRSP relied on the SODECOTON extension system for both on-farm
research and for general extension of the improved technologies (Table 4.5).
W

Annual costs for the nine-year period associated with the development and
extension of the improved cowpea technology are summarized in Table 4.6. The last

column, I"I‘otal Annual Costs” is the cost stream used in the benefit-cost analysis.

49

Table 4.5 Estimated Total Expenditures (Nominal
SUS), SODECOTON Extension Costs for
Cowpea Technologies, Far North Province,
Cameroon, 1981 to 1987.

 

 

 

 

 

 

 

 

 

 

 

 

 

Year Annual costs for Total Costs
extension staff,
food and cash crops

1981 2,552,541 15,317

1982 2,461,364 14,769

1983 1,839,374 11,035

1984 1,858,072 11,147

1985 2,539,872 15,239

1986 2,507,110 15,043

1987 2,614,626 15,688

Total 16,372,959 98,238
Source: Estimates based on internal documents

supplied by SODECOTON accounting office,
1991.

 

50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.6 Estimated Total Costs (Nominal SUS) for COWpea Research and
Extension Programs, F ar North Province, Cameroon, 1979 to 1987.
Year SAFGRAD CRSP NCRE IRA SODE- Total
J.P. 31 COTON Annual
Costs
1979 19,901 0 0 10,771 0 30,700
1980 17,325 0 0 11,001 0 28,300
1981 15,187 0 0 13,278 15,317 43,800
1982 15,657 131,565 0 33,478 14,769 195,500
1983 16,505 278,689 0 41,847 11,035 348,100
1984 34,916 332,003 0 58,585 11,147 436,700
1985 38,302 298,535 0 55,103 15,239 407,200
1986 32,638 272,893 4,890 26,462 15,043 351,900
1987 26,974 186,452 9,780 84,954 15,688 323,800
Totals 217,405 1,500,137 14,660 335,479 98,238 2,165,919
Source: Tables 4.1 to 4.5

 

51
4.2.3 Cowpea Benefit Stream

2 . . l B E E;

To estimate the benefits of an improved technology, three time series of data are
needed: (1) the annual market value of production (price times total amount of output
produced) since the introduction of the technology; (2) the annual market value of
production for this same time period, albeit conjectural, had the technoloy never been
developed and extended; and, (3) the annual on-farm costs of adopting the technology.

Specific data needed to estimate the mean benefit stream were: (1) yields
under three different sets of farming practices (total adoption of the cowpea package,
partial adoption of the package, and traditional practices)”; (2) corresponding adoption
rates of the new technologies, including adoption ceilings and the life span of the
technology; (3) total area harvested; and, (4) annual input and output prices.

E 1:. l 1 - E .

Numerous sources report cOWpea yields for one or more types of farming
practices (trial data and annual reports by the SAFGRAD and NCRE projects, the
CRSP, SODECOTON and MINAGRI). Reported yields (Table 4.7) range from 3,158
kg/ha, representing an upper limit achieved in on-station trials, to a low of 10 kg/ha,
which was one farmer’s response as reported in a survey of cowpea farmers (Ta’Ama,
1984; Kitch, 1990, respectively).

 

”Estimating the benefit stream requires estimates of W yields (under each
farming practice) for grain, leaves for food, and forage for animal feed. With traditional
practices, yield data were also needed for intercropped sorghum (grain and stover).

52

 

 

 

 

 

 

 

 

 

 

 

 

 

8.: c .55 as... 83.8
v5. “mm 833.8.» up: on gob—2:— EN . c: 83 mmoz\Q<~—0m<m\<m~
so... a: at. 88.36
i. 83.88.... 8.088... n o. a 8.8882 . 8mm . «3 a... .58... 5.0235. 058?...
meueu—n n3... aEcecba
8 8.8 Q... 8888... a? a 5.... 88.22... am . S 8. .3. .8... 8......8 :96
88 xi. 8...... .8... 888.8
8.8.... 8 8.8 as. 6.888... 8 8.888: «a. . 5 m8. 823.-.... .85
nib memeooba at. £838.
.88.. 888:8. 8888... 88.8882 8: - «8 3.8... 8.28.8 page
”889? mama—no.8 =a .65 guano? 8m . 3 83 aute... .258 £de
32a x85 a: 33638 :88:ch
v} gouge? 3389.5 n S N .eonecbceoz 8- . ck S. .3. .32 5.5.5 ZBOUmn—Om
28.»... 8.8.8 8 .25 8.2.8.3... «8. . .8 83.8. 8.80 1839..
3:9:
529$ ”5&ch one; 23> potency— §> 85cm 3.5

 

.8... 2 8... 8.850 .858... 582 a... 82:8 8......» .8853 2...» 8980

h... 03...“.

 

53

Given the range of estimates and their aggregated nature, merging data from
these sources is not a straight-forward exercise. For example, the Agricultural Census
and the CRSP farmer survey data are not difl'erentiated by variety or farming practice.
In contrast, SODECOTON yields, on-station CRSP results, and IRA/TLU estimates are
all associated with specific farming practices and one of the three extended varieties-
TVX 3236, BRl, or BR2. The high yields reported in the CRSP on-station varietal
screening trials are for monocropped cowpea, with insecticide treatments and intensive
management. Both the SODECOTON yields and the IRA/TLU on-farm trial data
represent production under the strict protocol dictated by the research and extension
agencies. The Agricultural Census data are aggregations of crop cuts and sample data
collected from representative farms, where annual yields are estimated from the reported
total production divided by the reported area harvested.

Yields used in the benefit stream are derived from the data described above and
insights gained through interviews of key informants. Yields for improved cOWpea
varieties cultivated in accordance to the complete extension package (monocrop,
insecticide, seed treatment, etc.) are estimated at 1,000 kg/ ha. Data explicitly indicating
that the yield for this package is 1,000 kg/ha do not exist, although farmers enrolled in
the SODECOTON extension program reported yields of 770 to 1,200 kg/ha (1985 / 86
Annual Report, SODECOTON). Cowpea researchers in Cameroon are uncomfortable
suggesting average yields. Responses during key informant interviews ranged from
“yields with the package will outperform local varieties by 200 to 300 kg/ha" to "if the
entire package is adopted, a good farmer will have yields of 1,500 to 2,000 kg/ha.” Given
that Kitch’s farmer survey (1990) indicated farmer yields no higher than 800 kg/ha, a

relatively conservative estimate (1,000 kg/ha) was set as the study’s “best guess”.

54
Yields for improved cowpea varieties that are monocropped, but not treated with

insecticide, are estimated at 400 kg/ha. Although several key informants contended that
'there would be no net gain in yield from adopting the new varieties over local varieties
unless insecticide is used,“ others argue that some gain would be possible with the shift in
management practices. Researchers have noted that when farmers grow improved
varieties, they almost universally grow them in monoculture, planted in rows in 1/4 ha
plots. The shift in farm management practices should result in higher yields per hectare, .
due in part to better weed control. Further, since during screening, researchers selected
improved varieties primarily for their grain-yield potential, they should outyield many
local cultivars that are grown for both their grain and forage production. This parameter
is conservatively estimated at 50 kg over the intercropped yield (in terms of pure stand
equivalents).

The pure stand equivalent yields for traditional mpea varieties intercropped
with sorghum are estimated at 350 kg/ha Several sources of information were used to
derive this estimate. One source was interviews of researchers, who estimated yields
under traditional practices to be 200 kg/ha or less. However, these figures appear to
have represented actual yields and not monocropped equivalents. Research trial data
indicate that monocropped equivalent yields for intercropped W are approximately
100 to 500 kg/ha, although the range is negatively skewed. Also, Kamuanga (1991), in
estimating production costs for traditional cowpea farmers, set cOWpea yields at 350
kg/ha. Finally, Agricultural Census data estimated average yields of nearly 600 kg/ ha
over the six-year census period (1984-89). From this range of values, yield estimates of
350 kg/ha are used in the base run, and then a range of yields are tested in the

sensitivity analysis section.

55
W

Although the CRSP has documented the importance of leaf consumption in the
traditional diet of northern Cameroonians (Ta’Ama 1984, Wolfson 1989, Kitch 1990),
there exist no empirical estimates of the quantities of leaf production from traditional or
improved varieties. .

While research in other countries" indicates that mpea can yield up to two
tons (wet matter) of leaves per hectare without reducing seed yield, lower leaf yields are
set for this study, based on the following assumptions. To estimate leaf yields, this study
assumes that leaf harvesting is limited to 30 to 40 days of the crop’s eighty-day growth
cycle. During this forty day period, if total leaf harvest is set at one ton per hectare,
then leaf yield is approximately 25 kg of leaves per day. An average household grows
approximately 1/2 ha of cowpea”, implying a daily harvest of 12.5 kg (27.5 lbs). This
study considers this level of harvesting and consumption to be unrealistic. While leaves
are sold in northern Cameroon, key informants contend that there is a thin, intermittent
market which would discourage harvesting the maximum possible amount of leaves.
Hence, a lower value of leaf production is set at 300 kg/ha for monocropped improved
varieties grown without insecticides and 400 kg/ha for intercropped traditional varieties.
A higher leaf yield for traditional varieties is based on two assumptions—traditional

varieties are grown specifically for leaf production, and the leaf flavor and/ or texture of

 

1nIrnungi and Potter quote work by Oomen and Grubben. Other sources reporting
leaf production include Mehta for Uganda, Martin for Senegal, and Connifi' for
Botswana. .

22This figure is based on the 1990 CRSP survey which estimated that the average
amount of land in cowpea production per farming household is 0.62 ha (Kitch, 1990).
This figure is biased upward since the survey’s sample frame consisted of "important”
cowpea farmers in villages in the principal cOWpea cropping zone.

56

traditional varieties is preferred over improved varieties since the latter were selected
only for grain yield potential When insecticides are used, leaf yields are set at zero
since farmers rarely harvest leaves from fields sprayed with insecticide due to chemical
residue. A range of yields are assumed during sensitivity analysis and these results are
reported below.

WW

Farmers produce forage (cOWpea hay) by uprooting and drying the entire cowpea
plant once grain pod harvesting is complete. Farmers store this hay and then either feed
it to their livestock or sell it during the dry season.

In 1989, the Maroua IRA/TLU unit measured cowpea hay production in an on-
farm trial conducted over 9 sites. Yields over all sites for two of the extended varieties
were reported to average 833 kg/ha for BRl and 607 kg/ha for TVX 3236 (NCRE 1989
Annual Report). It is unclear whether the yields are for dry or wet matter, although this
analysis assumes that the data are for dry matter. Given that these data are based on
researcher-managed on-farm trials, on-farm yields could be lower. Yet, improved
varieties were not selected for their forage production traits, implying that, genetically,
traditional varieties probably producer higher forage yields. Also, in the field, the
primary insect damage is to the plant’s flowers and pods, not leaves. Insect-damaged
plants tend to compensate for the loss of flowers/pods with more leaf production,
implying greater forage production when insecticides are not used. Based on these three
points, forage yields (monocropped equivalents) are assumed to be 750 kg/ha for
intercropped traditional varieties, 725 kg/ ha for monocropped, but not sprayed,

improved varieties, and 700 kg/ha for insecticide-treated, monocropped improved varieties.

57
E I 1:. I I _ E . ! 5 t
. Both the CRSP and the IRA/TLU conducted trials for intercropped cowpea and
sorghum. Again, results are difficult to compare due to different protocols used across
trials. Ranges of sorghum yields reported by researchers are presented in Table 4.8.

As noted by key informants, farmers seldom adopt new varieties and chemical
treatments (eg., insecticide applications for mpea and urea for sorghum) without also
converting from an intercropped to a monocropped farming practice. Hence, the
IRA/TLU on-farm trial data when no chemicals are used are most indicative of actual
farm yields for intercropped sorghum. Further, W is generally grown in drier
conditions where sorghum yields will be lower than average, suggesting that intercropped
sorghum yields will be below the province-wide average monocropped yield. Although
the base run for the analysis assumes a monocropped equivalent sorghum yield of 600
kg/ha, a range of yields are tested during sensitivity analysis and these results are
reported below.

Stover yields are not available, although sorghum stalks have a variety of uses in
northern Cameroon (eg., animal feed, fencing, roofing, raw material for toys). Minimal
values (5 fcfa/kg market value, 200 kg/ ha yield) are assumed in recognition that
sorghum stalks have some value and that this value needs to be included in the benefit
stream. In the absence of reliable data, sensitivity analysis is used to test the importance

of this parameter.

58

 

25.5....

 

 

 

 

 

 

 

 

 

 

 

 

.o 8.2. N cod—.20. .28 830.802.. 0}» a 2...... 2.5.8....
.5... 8.38 .5... 8.8.8.5 .2... 18...... .8 8... . 8. 8. .8. .82 ......s........ .82.
8...». c. .8... . 8.2... 58......
8 m._ 8. a? .8 .5... .<>> .5... 88.8.25 8... 8.... . 2.... 82 22023.53:
8..» a .8538. o: 820...... . .3... 58......
8.88 o... .5... 8.8.225 .8. 85....» 8. - .3. 82 $023.55..
2»... 5 v0.80» 2.08.8... 000... 8.2.. 58......
.8... 2:... cm 5... .8... 85....» 88888.... 88.. - 8... $82 802355.225:
8...— ... 25.5.... ....0E...0... 2...... 88.-....
88 .8... 23.. 8 .5... .88....» 1.8.5.8.... we... - 2. 8.82 502356825:
820...... :88... 2...... 8.25.8 5.8%....—
...... 83.2.... 252.8. 15.5.8. .8. 88.5.»...— 8n.. - one 5... 28.-.... ZOHOUmQOm
035.102.2008 .83... 5 v2.00» 2.0.58... V00» 8.25.8 8.62.9...
.258... m.— cm 5.3 .nmm bot? 60.520252 8a.. - 8. 3-2. .5 .23 88.-.... ZOHOUMQOm
8.2.... .588... .... 8.... 8:828... 8.... - 8. 8.82 888 88.8.8...
.23...
8...... .58.... 8...... .8... 88...... as... 88.... s...

 

.82 2 82 888...... 8......2 «8...... 88> .5... 25.. 88......

a... 05m...

 

59
W

Adoption is conceptualized as a cumulative process, increasing over time and
finally reaching a ceiling. Since estimates of adoption were not available for all years of
the analysis, available estimates for selected years were used as parameters to estimate
annual adoption rates. For the improved cowpea package, point adoption rates were
derived from IRA and SODECOTON documents and two surveys-both conducted in
1990—of farmers in northern Cameroon. To estimate the adoption ceiling and the
annual adoption rates for the years prior to and after 1990, a logistic functional form was
assumed for the adoption / difiusion pattern.

The first survey, reported by the CRSP, indicated that 25 % of cowma farmers
sampled grew improved varieties. The second survey, reported by IRA/TLU/NCRE,
indicated that the 1990 adoption rates for improved cOWpea varieties were 12.6, 26.6,
29.6, and 39% of all cotton farmers for the Guider, Tchatibali, Kaélé, and Maroua
SODECOTON extension zones, respectively.

The best indicator of farmer adoption of the complete package is the use of
insecticides. Kitch reported that 34% of the farmers surveyed ”used insecticides
regularly“ while the TLU reported that only 10% of the farmers surveyed “used
insecticides on cowpea". Since Kitch’s survey did not determine usage of insecticides by
commodities, the 34% overestimates insecticide use on cowpea. Conversely, the TLU
surveyed cotton farmers, a target population which excludes several large cowpea

growing areas. Hence, the TLU’s 10% probably underestimates insecticide usage by

cowpea farmers.

60

Given these data, a 25% adoption rate of the improved package is assumed for
1990 in the base run of the analysis and is used as a parameter in the logistic function
used for estimating annual adoption rates (Equation 4.2).

Using these surveys to estimate adoption rates had two limitations. First, the
sample frames used for the two surveys may not be truly representative of mpea
farmers in the Far North Province. The CRSP survey purposively targeted villages from
the principal cowpea production areas and I'important" cOWpea farmers within these
areas, who were administered a brief, sixteen question survey on ma production and
its constraints. The second survey was based on interviews of 1,003 SODECOTON
farmers throughout the region. Surveyed farmers were selected at random from '
SODECOTON producer lists, considered the most complete sample frame available in
the region. However, the cotton and W production zones do not completely
overlap, suggesting that many cOWpea farmers were excluded from the interview process.

A second constraint in interpreting the results of the two surveys is that both
collected only qualitative data on adoption of new varieties-whether or not new varieties .
were grown by the interviewee. Data as to actual area in production of new varieties
was not solicited during the interview, nor was the production mix of area in the new
package versus traditional practices.

These various data sources were combined to estimate the adoption pattern as
described below. Initial adoption figures for 1984 and 1985 are estimated from
SODECOTON production figures and IRA/SAFGRAD/CRSP on-farm trial data. In
1984, SODECOTON extended the then experimental cOWpea package on 11.25 ha.
Given that the package protocol required that each plot be 1/4 ha, it is assumed that 45

farmers participated. That same year, IRA/SAFGRAD/CRSP conducted on-farm trials

61
at 28 sites, using the same protocol as SODECOTON for a total of 73 participating

farmers. In 1985, SODECOTON reported 81 ha in production, implying 324 farmers
'adopting', while IRA conducted on-farm trials at 24 sites—for a total of 348 farmers
”adopting“ the package in the second year it was extended.

Since the two surveys in 1990 only report the percentage of farmers adopting the
new variety, initial adoption figures (1984 / 85 data) need to be converted to the same
base units (percent of farmers, not number of farmers). Assuming that the average
cowpea area per farming household is 1/2 ha”, the number of cowpea farming
households is estimated by dividing the cowpea area by 1/2 ha. For 1984, the
Agricultural Census reported that 23,470 ha were harvested, implying 46,940 farming
households. Dividing 73 (Le. number of adopters of the complete package) by 46,940
leads to an estimated adoption rate of 0.2% for the first year the technology was
extended. Similarly, for 1985, the census reported that 30,232 ha were harvested,
implying 60,464 households growing cowpea and an adoption rate of 0.6% for the second
year.

Using the estimated percentage of farmers adopting in 1985 and in 1990, the
parameters for a logistic function (Equation 4.3) were eulculated by solving the following
two-variable, twooequation algebraic problem.

0.6 + (0.6 t be '2) = K (42)
250+ (25.0 t be") = K

 

”Kitch’s findings are reported above. The Agricultural Census also reported, by
stratum, the number of farms and area planted per farm for 1984: 130,000 farms planted
less than 0.25 ha cowpea, averaging less than 0.1 ha; 14,000 farmers planted between 0.25
and 0.5 ha of cowma, averaging 0.4 ha; and, 12,000 farmers planted more than 0.5 ha,
averaging 1.1 ha.

62
The values for K and beta are 35.2 and 444.1, respectively, where K is the adoption

ceiling (i.e., the maximum percentage of farmers adopting the complete cowpea
package). These parameters are substituted into the following logistic equation to

project annual adoption rates.

P(t) = (4.3)

1+be"

where:
P(t) 2 Cumulative total of farmers adopting technology, period t
K = Adoption ceiling, a parameter of the logistic function

b = A parameter of the function
t s Time period, number of years since extension of technology

From equations 4.2 and 4.3, annual adoption rates were projected over a
thirteen-year period (1986 to 1998), while the adoption rates for 1984 and 1985 were
based on IRA and SODECOTON production figures (Table 4.9).

Compea researchers in northern Cameroon have noted that farmers do not adopt
the improved package on all of their land in cowpea production". Although there
exists no data on the shares of improved vs. traditional cowPea, estimates used in this
study were based on the area planted per farm and on the extension protocols for
mpea production. Since, on average, cOWpea farmers grow approximately 1/2 ha of
ma, and the protocol for the extended package requires 1/4 ha of land, the base run
assumes that the initial production mix is equally divided between new and traditional

varieties. As farmers gain confidence in the new technology, it is assumed that the

 

3‘ Some land will remain in traditional varieties due to the preferred tastes and
texture of leaves and grain of these varieties.

63 -
production mix will gradually shift so that by 1994 1/2 ha is planted in the new package
and a small residual (1 / 8 ha garden plot) is kept in traditional varieties, implying a long
term production mix of 80% new, 20% traditional (Table 4.9).

So far, discussion on the adoption rate has assumed total adoption of the
extended package—new varieties, insecticide sprayings, improved farm management
practices, etc. However, there is considerable anecdotal evidence that some farmers
adopt the new variety and farming practices (monocropped blocks of cowpea), but do not
use insecticides. Since pod yield gains are small and leaf gains negative with partial
adoption, the adoption rate will probably be low. Although the improved varieties were
introduced in 1984, difiusion of these varieties beyond areas under the direct influence
of SODECOTON and IRA is assumed to begin in 1986. Partial adoption of the package
represents an additional impact of the improved technologies (especially improved
varieties) and more benefits from research. The partial adoption rate is assumed to be a
fixed proportion (15 %) of the adoption rate for the complete package”, since the total
number of farmers exposed to the new varieties will increase as more farmers adopt the
complete package. It is also assumed that the use of new varieties without insecticide
represents a smaller share (25 %) of the farmers’ production mix of new and old
varieties/practices in their cowpea fields (Table 4.9).

W
Inputs used for cowpea production include seed, seed treatment, labor, and

insecticide. However, for this analysis, the key point is not the actual cost of inputs,

 

”Survey data indiested that 38% of the farmers were growing cOWpea m
monoculture, but only 34% were regularly using insecticides (Kitch, 1990). The fixed
proportion is based on these data and comments from interviews of key informants.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 mm Wm «.3 on «.mm 33
m4 mm fin Nun cw Nam 33
2 mm fin Nun 8 «.mm 32
3 mm mm flan cm “.9” 32
mg nu Wm 3“ on 9mm 32
m._ nu Nm 9.3 D. new. 82
2 mm 9n 9.8 on van ~22
«A mu 3. 3: no v.2.” 83
3 mm Qm fimg on <3 83
ed mm Wu «5 mm mg 33
3 R 2 3 M an 3 32
«6 na ad ad an ed 52
mod 3 «6 ad cm 2 owe—
a: a: c M... an ad mm...—
a: a: c fio cm No $2

as 8:25 $3 assess

owe—own Etna 05 Eco owes—one 829:8 Beeoefl

Rev 9 338:“ «5.83. 3% 2 838:“ $.83.

95. 8am tea—no.5 «09,50 Eugen 2.0 23. new—e95 «0930 wee—Emu
comic—at. £2 .83on we gun—=55 eozmou< i=2 beaten we 02338—5 30>

.weaéwa E3? £082.30 .8525— 552 Sn 532250.“. 3E0 ecu 88“ comic—é 3.3me ed. 035—.

 

65
rather whether expenditures on inputs at the farm level changed with the adoption of the

package attended to farmers.

TheopportunitycostofcowPeaseedisrelativelyhighsincetheannualpeakin
cOWpea market price coincides with planting time (late June/early July). However, no
documentation was found which indicated that improved varieties command a premium
price in the market over traditional varieties, a view confirmed in interviews with key
informants. If this is the case, then seed costs can be left out of the benefit stream.

On the other hand, the market may fail to reflect actual costs of seed production.
Seeds for improved varieties were probably more costly to produce, at least in the initial
stages of diffusion since they were produced by either IRA or the NCSM project. Since
cowpea is self-pollinated, varieties can be propagated at the farm level without a -
supporting seed industry. This implies that after the initial distribution of improved
seed, seed costs do not differ between new and traditional varieties. Again, seed costs,
with the possible exception of the coax of the “starter“ seed, can be left out of the
benefit stream.

Further, on-farm seed costs may be lower when the improved package is adopted.
Although extension recommendations for seeding densities for improved varieties are
greater than traditional seeding densities (20 verses 10 kg/ha), fields of monocropped
mpea require no sorghum seed.

In the aggregate, difl'erences in the cost of seeds between traditional and new
farming practices are assumed to be minor, relative to the magnitude of the cost and
benefit streams of the analysis. Hence, seed costs of the improved package are not

included in the base run of the analysis.

66 _
Chemical seed treatment, usually thioral, is part of the extension

recommendation, but is not usually applied on traditional cowpea varieties. Although
seed treatments are usually provided at zero to minimal cost to farmers by either
SODECOTON, IRA, or MINAGRI, their distribution is a cost to these institutions. In
1991, SODECOTON reported that thioral, delivered to Garoua, costs 1,295 fcfa/kg
(SODECOTON internal document, 1991). Extension agencies distribute seed treatments
in prepackaged 50-gram packets. The extension package recommends that each of these
small bags of thioral will treat 5 kg of seed, the amount needed for a 1/4 ha plot. These
costs are included in the analysis (Table 4.10).

Insecticide cost estimates are based on an aggregation of data available on
several general purpose insecticides distributed by SODECOTON“. Costs (quoted in
fcfa per liter) for each of these insecticides were very similar across types of insecticides
and for the years when prices were available (1987-1991). The overall average price of
3,200 fcfa/liter is used as the base cost of insecticide for the analysis (Table 4.10).
Application rates were uniform for each type of insecticide-1 liter/ha per spraying with
two sprayings per growing season. Investments in insecticide spraying equipment are
considered sunk costs incurred by present or past cotton farmers and are not included in
the benefit-cost analysis. Throughout the region, spraying equipment has been
distributed to cotton farmers for over thirty years. Given that this equipment is already
available to farmers, it is assumed that no additional costs for sprayers are incurred

when adopting the cowpea technology.

 

”These insecticides include monocrotophos 300, Nurelle-Dursban 30-280, Nurelle-
Dursban 30-450, cypermethrine-monocrotophos 30-300, Fenom N324, and Cymbush.

67

.82: 09.228 ecu ”5:83.:— mE— ”8:05.39 3535 ZOFOUNDOw ”850m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.8 8.8 88.8 8... 83 82
8.8 8.8 88.8 86 82 82
8.8 8.8 8% 8... 83 82
8.8 8.8 88.8 85 82 82
8.8 8.8 88.8 n 83 88.: 82
8.8 8.8 88.8 85 82 8.:
8.8 8.8 88.8 86 83 83
8.8 8.8 8% 8.: 83 82
8.8 8.8 88.8 85 82 82
8.8 8.8 88.8 8... 8.8.: $2
8.8 3.8 88.8 8.: 82 82
:8 8.8 88..” 8.9 83 82
8.2 8.2 88.... 86 82 82
8.: 8.: 88.8 85 82 82
8.2 8.: 88.8 and 82 83
35 A25
58%.... .o 855% 8 cease:
380 39: 5288.5 02882: 35 9.30:
Sheen. 3:52 «.0950 aomBoU 20:53.5. coon “5.53;. poem
3:036ch :33. a: com H80 300 is hem a: com «80 #60 3:3 com 50>
.88: 2 8.: £83an .858: 532 as $5.08
«838 365.5 2: 3.83.. .23 3383. see a: 53.85 8&8 as .85 gas as 2.3

 

68
Labor costs are not included in the analysis. Although IRA reports the

production costs/partial farm budget for cowpea production (Kamuanga, 1991), a
comparative study between new and traditional practices was not available. Speculative
differences include lower weeding costs with monocropped cowpea, additional labor
needs for insecticide application in monocropped cowpea, and different harvest labor
needs for the two practices-monocropped cowpea requires more time to harvest COWpea
grain due to higher yields, but require no effort to harvest leaves or sorghum, labor
demands in the traditional intercropped system.
W

The analysis quantifies the economic value of the five outputs generated by the
three cropping systems employed by cowpea farmers”. Time series of prices for each
of the outputs are generated from data when possible and from comments by key
informants when data are insufficient. Established markets exist only for cowpea and
sorghum grain. 'Ihin markets with intermittent sales exist for cowpea leaves (for food),
and for sorghum and cowpea hay, although no records were available on the historic
prices of these three products.

MINAGRI reported monthly retail prices for cowpea and sorghum grains over a
six-year period (1985 to 1990) for five ”urban” markets in the Far North Province”.
However, these series are incomplete, with as many as six months of price data

unreported for a given year and market. Further, sorghum prices are disaggregated into

 

2"The five outputs are cOWpea grain, leaves, and forage and sorghum grain and
stover. The three cropping systems are traditional intercropping of cowpea and sorghum,
complete adoption of the improved technology, and partial adoption.

”The five cities are the prefectures (equivalent to county seats in the United States)
for the Far North Province: Kaélé, Maroua, Mokolo, Mora, and Yagoua.

69

two time series, one each for white and red sorghum. The TLU provided a second
source of prices, reporting average monthly prices across six rural markets. These
averages, based on two years of price data (1989-90), ranged from a low of 101 fcfa/kg in
December to a high of 196 fcfa/kg in August. ’

Although these two data sets indicate that seasonality and annual rainfall greatly
influence the market price, the series are too incomplete to be used to calculate a
representative price for the analysis. The base run price is an average price (calculated
from MINAGRI data) for the five markets at harvest time (November) in a year (1988)
of “normal” rainfall A second assumption set the market share of white and red
sorghum at 50% each; thus, the mean price of the two was used during the W
analysis. A range of prices is then tested during sensitivity analysis.

Prices for cowpea leaves, and cowpea and sorghum stover were based on
comments from key informants. Although these outputs are sporadically marketed, they
do have economic value within the farmers’ households. Since these prices are based on
qualitative data, they are tested during sensitivity analysis. Price data for outputs are
summarized below (Table 4.11). Prices are reported in fcfa but are converted to $US in
the analysis.

70

Table 4.11 Estimated Average Annual Market Prices for
Cowpea and Sorghum By-Products, Far North
Province, Cameroon, 1984-1998

 

 

 

 

 

 

 

 

 

 

By-Products , Market Price
_ fcfa / kg
Cowpea grain ' 155
COWpea leaves 35
Cowpea forage 25
Sorghum grain 60
Sorghum stover 5
Source: Estimates based on MINAGRI and TLU data and

interviews of key informants

4.2.4 Base Run for Cowpea ROR

W
To conduct the analysis, annual total harvested area was disaggregated into three

subsets: area in traditional practices, area under complete adoption of the improved
package, and area where only the new variety (and not the complete package) had been
adopted. The relative shares of each of these cropping systems are calculated by
multiplying the adoption rates times total cowpea area harvested. For the period 1984 to
1989, total areas harvested are based on Agricultural Census data reported in Table 2.3.
The six-year mean of these data were then used as the annual total area harvested for
the period 1990 to 1998.
Mam

For each of the cropping systems, production totals are estimated based on the
yield data reported above. Estimated per hectare yields are multiplied by area in

production per cropping system to give total production figures for each system. The

71
production figures for the three systems are then aggregated to determine the total

annual production, given the introduction of the extension package”. A second total
production figure is estimated for the region, based on the assumption that the improved
package had never been developed and extended. It is calculated by multiplying
estimated total area harvested by the estimated per hectare yields for the traditional
system (i.e., local cowpea and sorghum intercrop). The difierence between these two
totals represents the gain in production due to the adoption of the improved package”.
The market value of this gain in production is then estimated, using the price data
discussed above.
law
For the two ”adopted” cropping systems, aggregate input costs are calculated by

multiplying the total area in production times the total per hectare cost of inputs. Per
hectare input costs for the adoption of the complete package include both the cost of
seed treatment and the cost of insecticide sprayings. Input costs for partial adoption
only include the cost of seed treatment. These aggregated input costs are then
subtracted from the benefit stream.
W

. Gross benefits are determined by summing the market value of the gains in
production, minus the increases in input costs incurred by farmers who adopt the

package For this analysis, gains and costs are reported in SUS (Table 4.12).

 

”F or example, in 1989, total production of cowpea grain is estimated to equal
(16,988 ha " 9.2% " 1,000 kg/ha) + (16,988'ha ‘ 0.6% ‘ 400 kg/ha) + (16,988 ha "‘
90.2% " 350 kg/ha) = 6,970 metric tons of mm.

30For example, in 1989, the estimated gain in W grain production equals 6,970 -
5,950 =- 1,020 metric tons.

72
The time horizon of the benefit stream is fifteen years, beginning in 1984, the

first year TVX 3236 was extended by SODECOTON to farmers. By 1991, the year of
this analysis, the improved varieties and farming technologies were experiencing a
relatively high level of adoption (estimated to be 25%, as noted above). Key informants
contend that cowpea is becoming an alternative cash crop to cotton in many of the more
drought-prone areas of the Far North Province. Hence, projections of the improved

varieties being grown for seven more years appears quite plausible.

 

The internal rate of return for the base run is 15.5%, calculated for the net

benefit-cost flow reported below (Table 4.13).

73

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m3 SH nu- So- 2”. new- 33 wee"
n8 G~ mu- Se- 3. men. v38 33
mno “2 ma. Se- Hm- N3”- v38 33
c3 3_ ma. woe. 3. Sn. >98 33
So o2 3. 8o- 2”. own- 2:8 33
New 3— MN- emw- an- 2 m- 33 mee—
mcw «.2 “N- Eh- om- mwm- omen «ea
:8 8: 8. c8- 8- n8- 3: 8.:
Sn 5 3. we... 2. 9:. M: 2 83
mm: mm m- ow? e. or- e3 awe:
v: on n- :7 e- 8. m3. $3
9. e T av. 7 2. a: 52
“N m 7 cm. 7 5. mm owe—
» fl o- h- o- m- on 32
N o c- N- c- T n 53
owe—own— umoU 5:335 couuzuocm 533—5.:— =ot3vo£ eomuoavoi
.0323: 3%: 0305 EEO 380m 304 £80
an E0830 :5:wa :5:wa «09.60 «3300 «0360
mauoeom 8:518 we 38> mo 38> ac 38> we 38> E :80
9.90 88H couscom 30.60% coca—80m c00=v0m no 33> 30>
.3: 2 88 £88an 8588 582 5.
633—3509 «3300 @3995 no coma—.05 tea 3332309 05 Bob £3» 89 5 .0538 395 «3. 03¢.

 

74

Table 4.13 Estimated Benefit-Cost Flows (in ’000 $US) for the Improved Cowpea
Package, Far North Province, Cameroon, 1979 to 1998.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year ' Gross Benefits Gross Costs of Net
from Cowpea Research & Benefit
Package Extension - Flow
1979 0 -31 -31
1980 0 -28 -28
1981 0 -44 -44
1982 0 -195 -195
1983 0 -348 -348
1984 2 -437 -434
1985 8 -407 -399
1986 21 -352 ' -331
1987 43 -324 -281
1988 174 0 174
1989 195 0 195
1990 517 0 517
1991 744 0 744
1992 803 0 803
1993 892 0 892
1994 967 0 967
1995 970 0 970
1996 973 0 973
1997 973 O 973
1998 973 0 973

 

 

 

 

75
4.2.5 Sensitivity Analysis - Cowpea

Although the base run is the best-judgment estimate of the returns to W
research and extension in northern Cameroon, sensitivity analysis is conducted to test the
robustness of that estimate. Further, given that some data used in the analysis are
estimates based on informed assumptions and not actual empirical findings, sensitivity
analysis indicates how each assumption affects the results.

Over sixty additional estimates of the IRR to cowpea research and extension
have been calculated, modifying the values of one or more of the model
parameters/variables for each of the 60 runs. From this analysis, eight
parameters/variables were identified as having a significant influence on the estimated
rate of return (Table 4.14). Other parameters/variables, whose values were varied,
changed the IRR by 20% or less from its base run value and hence, these results are not
reported. The thirteen runs in which the rate of return was affected by more than 20%
are discussed individually.

76

.06 “and. a. vow—RF. :32: 88 .35.- on. be 8583- ..- a 88 .88..

.833 a a: a... o... a. 95.8 8:83 2F .3 93.... a. 8.5.2 a. a... 8:8... .352.

.3 892 x. 3.8:... B... one»: :5 name ficfln .83.: .555 .2 0.3 a 3:2? 3 3.82... .2. i=5...

031...: o... .5. 333 E: 33 8:80.52 .9.

 

ed +
ed.

tn-
EN +

he +
ONT

tn.
QN +

mam.
—.v+

«re +
Wan-

v.9 +
fie.

28:9 .0 38> E 33.55 :-
oSatg .«o 38> E onuuuuov -
"Ea. S: .0 3.5 5 99:5

 

mum—
Q:

fin—
fid—

—.N—
n.”—

od—
0.2

”.3
oi?

od—
We

03:5, .0 38> E 03965 :-
oSatg ..o 38> E omen-00v -
no. 33 5:

 

a 88 use See

 

Seances mac—Ea $3 8.1 cage—é

 

:5 3.35: SE

 

0.09

:85 £2.93

 

.9

£93 3938
"33$ 8....

 

nerd“:

E02: 5 £95 £2.93

 

gaua

ESE S E...» 3.58

 

$3M:

exec—M3

an: won-.85 68.5 33.50
A285 so.»

 

 

5 EJ—

 

o and

 

m =3—

 

v =3—

 

n =3—

 

N a...“

 

— =3—

 

ESM— g

 

33...; as.

 

82.22 .8236 .083:
532 an .8305 a .283. 39.8 a. no: 2.. a. 8.58 2892.5 2.. 8.3.5 $— 3 31> .532 {1.2 beans...

3.! 033—.

 

77

9...... 838 8.3.2.8

.Bs.§.§£§8§.5§s§§.&§...§.-§cfif..fi£§fixsfi§la§.38§z£§.§889
1. 8...... 88...... 38...... 8.. 3...... 2.. 8.9.8. 83%

o... a. 8:8». a. a... 5.92.: .5....... £38 33...... .5... «Sin 8...... 8.8:... .2. 89.8 .38. 8. a... 29.2.... .8. a... .5...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

n: + ad. —.n+ 0.9-18> ._o 38> E 83.6... =-
3..- «6+ 2.”. 0.9-v.9 .0 38> a. 3380.. -
fie. o.v+ 0.5+ 0.: ”ED... .5: .8 38> a. 5:25
o..." ha. ed. 03...: .o 38> a. 888... a.
n. v.2 h: 0339 .0 38> a. 83.60.. a
q... 2. a .8 3. 8. s... 5..
.8. u a 8... a a a 8n n a a 8... a 88 3.8 .89
.8. u a a a a a a 88 u a an... 989.... 5...... 8. 2... 8.88...
.8...
.8. u a a 9. .2. . 88.8 n a a 8...“ A... 8.32... .2...
a .. a a 88 + .. a a 8 a...» 22.98
.8. n .. a a a $8 + a a a an. a...» 89.8
“a..\_..... 8....
a .. a a 8n. - a a a 8. 98.2... a. 55 £2.98
a a a a $9 - n a a can nap—8... a. :8... 3960
.8. n a a a a 8n - a a .. 8... 9... 82...... .5... 89.8
".23.. 22>
.. .3. n. .3. N. .3. .. .5... e. .3. . 8... a .3. .3. a... 33...; .3.
88.9... .8285 .885... 582
8.. 8.8.3. d 8.83. 89.8 8. 4.9. o... a. 8.8.0 .8888 .3. 8.......> 5. .9 8....> .598: «1...... 8.8.88 2.8 .v... 2......

 

78

The analysis was most sensitive to one parameter—the cowpea grain yield for the
complete adoption of the improved technology package. Originally fixed at 1,000 kg/ha,
four alternative values are reported, indicating rates of return ranging from +24.8 to -
18.0 percent. In Run 1, the parameter was varied by plus or minus 10%, implying yields
of 1,100 and 900 kg/ha, respectively. Similarly, in Run 2, the yields varied from the base
run by 25 %, implying parameter values of 1,250 and 750 kg/ha. Although the
parameter’s value greatly affects the returns to research, key informants have a high
degree of confidence in the expected yield of the complete package. Hence, varying its
value by 25 % is probably excessive, and the resulting negative rate of return is unlikely.
Thus, Run 1 better represents the range of values proposed by key informants, indicating
a possible range in the IRR of 9.5 to 19.9%.
B 3’ E 1:. I l l. . I

When the value of the cowpea grain yield estimate for the traditional,
intercropped farming system was varied by plus or minus 25% (287.5 kg/ha), the IRR
ranged from 10.0 to 19.6, respectively. Since the traditional system is being replaced by
the improved technology, a decrease (increase) in the value of this parameter will result
in a higher (lower) rate of return. By setting a lower output value (263 kg/ ha) for the
”defender crop", the new technology will inevitably appear better than before. Although
the base run estimate for this parameter, set at 350 kg/ ha, is the best available, key
informants generally skewed their estimates of intercropped cowpea yield downward,
indicating that the base run, if wrong, more likely overstates the yield and

underestimates the IRR.

79
W

By varying the value of the yield parameter for intercropped sorghum by plus or
minus 25%, the IRR ranged from 12.1 to 18.3%, respectively. Again, by diminishing the
value of the output of the ”defender crop", the returns to adopting the new technology
are increased. However, estimates of intercropped sorghum yield are generally skewed
upward, indicating that the base run yield of 600 kg/ ha, if wrong, underestimates the
output of the traditional farming system and overestimates the rate of return. Even if
this is the case, a twenty-five percent increase in the parameter (to 750 kg/ ha), resulted
in only a 22 percent decrease in the rate of return, relative to the base run.

The rate of return estimate varied considerably when the cowpea price parameter
was varied by plus or minus 25 percent. As price was increased to 194 fcfa/kg, the rate
of return increased to 22.2%. As price fell to 116 fcfa/kg, the rate of return decreased
to 3.5%. Trends indicate that prices, if misspecified, are underestimated. Recently
developed (since 1987) improved storage technologies should allow farmers and grain
merchants to delay sales to capture higher post harvest market prices which occur during
the dry season. If this holds true, then the base run underestimates the returns to
research and extension.

Varying the sorghum price parameter had less dramatic effects on the estimated
IRR than similar variations in the cowpea price parameter. When sorghum price is set
at 45 fcfa/kg (75% of the base run value), the IRR is 18.3%. When the same price is set
at 75 fcfa/kg (125% of the base run value), the IRR becomes 12.1%. Again, the rate of

return decreases when there is an increase in the value of the crop which adopting

80
farmers no longer produce—in this case, intercropped sorghum. Although varying this

parameter’s value significantly affected the IRR, interpreting this result is extremely
difl'icult. Sorghum price, depending on the weather, can vary significantly from year to
year (Kamuanga, 1991, Johnson, 1987). Hence, the range of values for the rate of return
in this run simply indicates the most likely returns, assuming relatively stable prices in
the food-crop market.
W

Area harvested data are estimated from the published reports of DEAPA, the
Cameroonian agency responsible for the annual Agricultural Census. By testing the .
robustness of these data, the analysis determines to what degree the rate of return
estimate relies on the accuracy of this secondary data. When varied by plus or minus
25% of the reported values, the IRR estimate was 18.5 and 11.9% respectively. Given
the assumptions of the analysis, an increase (decrease) in the amount of land cropped to
cowpea leads to an increase (decrease) in the returns to the project.
W

The annual estimate of the percentage of farmers adopting the new technology
was varied by plus or minus 25%, resulting in IRRs of 18.6 and 11.7%, respectively. By
increasing (decreasing) the number of farmers adopting the new technology each year,
the returns to the project increase (decrease). The annual adoption rate is a difficult
parameter to estimate. Hence, this run indicates that although the analysis is sensitive to
this parameter’s value, a 25% decrease in the annual percentage of farmers adopting still

leads to a positive, albeit lower, rate of return to cowpea research and extension.

81

W

Total annual costs were varied by plus or minus 25%, resulting in lRRs of 12.7
and 19.4%, respectively. Although most com figures were directly quoted from project
and institution documents, there is still a possibility of incorrectly estimating cost
variables. The decision to include or exclude a given expenditure is not always clear cut.
Serendipity within a well-established research system, spill-over benefits from other
projects, and complementary research expenditures (eg., some training costs) are
examples of potential costs that may not be included in the cost stream. Hence, by
varying the cost variable, an estimate is made of how sensitive the IRR to W
research and extension is to higher or lower total costs. For this analysis, the IRR was
still favorable at 12.7%, even with a projected 25% increase in annual costs. ‘
W

This run projects a droughty future in which yields for all crops are 25 % lower
and prices are 50% higher. Given that W is more drought tolerant, the returns to
this scenario should be, and are, higher than for the base run—20.1% as compared to
15.5%. Run 10 is more than an academic exercise since much of the cowpea zone is in
an area which has been experiencing a decade long period of drier than normal weather.
With cowpea’s comparative advantage in such conditions, the base run may actually
underestimate the returns to W research and extension.
W

The two sources of data for total cOWpea area harvested are discussed in Chapter
2, as are the justifications for using DEAPA’s data over MINAGRI’s. However, for Run
11, MINAGRI’s data are substituted for DEAPA’s and a resulting IRR is calculated.

For Run 11, the IRR to cowpea research and extension is 22.5%, which is expected since

82
MINAGRI’s estimates of total cowpea area are consistently higher than DEAPA’s. As

with Run 7, as total area in W increases, so does the area where the new
technoloy has been adopted, and thus an increase in the returns to research and
extension.

Given some anecdotal evidence presented by key informants, the amount of land
grown to W may actually be increasing. If this true, than the assumption used in
the base rum-fixing the total cowpea area harvested at 23,600 ha for the years 1990 to
1998~underestimates the IRR. As storage constraints are addressed by the research and
extension system, and as cotton, an alternative cash crop, becomes less profitable due to
structural changes within SODECOTON, cOWpea may become a much more significant
crop in northern Cameroon. Run 12 tests this scenario by assuming that total area in
production in 1990 is 23,600 ha and that this area increases by 10% each year after 1990.
The resulting IR is 19.5%.

c e t

This run relaxes the assumption that Cameroon’s currency is not overvalued.
Given that some anecdotal evidence in 1991 indicated that the fcfa in Cameroon was
overvalued by approximately 40%, this run assumes that the overvaluation gradually
increased to this level during the 1980s. Starting in 1981, an annual 5% incremental
increase in the percentage of overvaluation is assumed (i.e., in 1981, the currency is
overvalued by 5%, in 1982, by 10%). Thus, a 40% overvaluation is reached in 1988,
which is then held constant for the remainder of the analysis (i.e., through 1998). The
shadow exchange rate is calculated by multiplying the market exchange rate by a

conversion factor (1 plus the foreign exchange premium, where the premium equals the

83
percent of overvaluation divided by 100). The shadow exchange rate is then used to

convert the values of all tradable goods within the cost and benefit streams to SUS. The
resulting IRR is 11.4%.
W

A best case/worst case scenario is tested in Run 14, whereby five of the key
variables are simultaneously modified by plus or minus 10%, resulting in IRR’s of 27.0
and 1.3%, respectively. This range in the IRR is. relatively academic since it is unlikely
that the base run misspecified all five of the parameters/variables in such a way that all
affect the IRR in the same way (negatively or positively). However, these results
indicate that the returns to W research and extension are most likely positive, and

potentially quite favorable-even under the most extreme set of assumptions.

4.3 Sorghum
4.3.1 Overview of Sorghum Technologies

Grain sorghum has been one of the primary foci of agricultural research in
northern Cameroon for over three decades. Early work (mid-1960s through the mid-
19703) conducted by IRAT and the SAFGRAD J.P. 26, a regional SAFGRAD project
preceding J.P. 31, included the collection and classification of local germplasm and the
screening of local varieties for desired traits. A short-lived breeding program was also
initiated in 1970. In 1974, IRAT terminated its work in Cameroon and in 1976 J.P. 26
came to a close, leaving only the Cameroonian government, through IRA, to fund
sorghum research. As a result, over the next several years sorghum research was limited
to maintaining germplasm and seed stock. In 1979, sorghum agronomy and varietal

screening trials were reinstated by the SAFGRAD Joint Project No. 31. In 1982, the

84
NCRE project greatly expanded sorghum research through the creation of a sorghum

breeding progam. In 1986, the NCRE project expanded its focus on sorghum,
establishing a sorghum agonomy progam in Maroua to complement the breeding '
research.

Throughout its history, sorghum research has focused on increasing gain yield,
given the production constraints of the region. In the mid-1980s, yield stability emerged
as a second research objective, as scientists recogiized that yield stability across a wide
range of environments and varied production constraints was as critical for meeting the
needs of farmers as higher yields.

The IRAT and SAP GRAD research programs identified several sorghum
varieties (IRAT 55, CE 99, E 35.1, and 38-3) for extension to farmers. However, these
varieties were never extended on a large scale, in part due to constraints in both seed
multiplication and extension resources. Not until 1986 were ”improved" sorghum
varieties (NCRE selected varieties S34 and S35) extended across large segnents of
northern Cameroon.

S-35 is unquestionably the sorghum research program’s most sigiificant
technological output. This variety, originating from India, is a short. cycle (90 day),
medium height (2.5 m), white-gained sorghum that has some resistance to disease and
insects. It was first gown in northern Cameroon in 1982 as one of several hundred
varieties screened by the IRA/NCRE sorghum breeding progam. From 1983 to 1986,
the variety was tested both on-station by the sorghum breeding and cereal agonomy
programs and on-farm as part of the SAF GRAD research progam. In 1985, the NCSM
project began multiplying S35 seed, producing 42 metric tons which, in 1986, much of
which was extended (purchased and resold to farmers) by SODECOTON in 1986.

4.3.2 Sorghum Cost Stream
E . . I Q | :

Two donor projects and two host country institutions contributed to the
development and extension of S35. SAFGRAD J .P. 31 conducted many of the early on-
farm trials that helped identify S35 as an appropriate “improved" variety for the region.
Within the NCRE project, the sorghum breeding unit in Maroua and the cereal
agronomy unit in Garoua contributed to the selection of $35. Given that IRA is the host
institution for the NCRE project, a portion of its expenditures also supported 835’s
development. As the only efiective extension agency in the region, SODECOTON also
contributed to the "success” of S35 both as a collaborating institution with on-farm trials
and as the principal conduit for extending the technology to farmers.

The cost . streams for the development of S35 extends from 1979, the first year of
the SAF GRAD project, to 1986, the first year SODECOTON recommended 835 to
farmers.

W

As stated in the discussion on cowpea costs (section 4.2.2), sorghum was one of
four food crops targeted by the SAP GRAD Joint Project No. 31, although it received a
disproportionately larger share of stafi" time and project resources. From 1979 to 1983,
SAF GRAD researchers allocated approximately 30% of their efforts to sorghum
research. From 1984 to 1987, that percentage increased to 60%, shifting emphasis away
from maize and millet. Since project costs are not disaggegated by commodities, these

percentage shares-based on interviews of key informants-are multiplied by total project

86

costs to estimate the relative share of costs for sorghum research". Otherwise, the
estimation of SAP GRAD cost streams for sorghum research follows the approach and
set of assumptions concerning detailed cost accounts that was applied to the cowpea cost
stream reported in Section 4.2.2. The resulting cost stream is reported in Table 4.15.,
SAFGRAD costs prior to 1982 were included in the cost stream, even though S35
was not tested in Cameroon until 1982. This is justified since much of the early
SAF GRAD work laid the foundation on which the later development of S35 depended.
From 1982 to 1986, SAP GRAD was responsible for testing 835 in its on-farm
research /pre-extension progam.
W
In the early 1980s, two research progams within the NCRE project allocated a
portion of their resources to sorghum research-«he sorghum and millet breeding unit
based in Maroua, and the cereal agonomy unit based in Garoua. Other NCRE
progams which focused on sorghum were established after 835 was initially extended,
including the TLU’s on-farm progam, and the Maroua-based sorghum and millet
agonomy unit. Since these latter two NCRE supported progams began after the

technology was extended, their costs are not included in this analysis.

 

31Estimates of each commodity’s share of total project costs are based on interviews
with Owen Gwathmey, Jerry Johnson, and Martin Fobassoathe three principal
researchers working on the SAFGRAD project.

87

vases—«>5 «0:. 3.6 895?. S...

A8353 wag—:2.» woo—330:“ nema— uo “moo rwov 3% he?» 05 E @0032: 8a Succum-

.=8=:o—. .n was deacon .2 Soc—5290 .O 53 goings ”3:08:89 838m ”venom

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mvnfinm mnfinmu wanna“ a: a: a: a: a: £80k.
finds 893 Saw: a: a: a: a: a: 3%
v8.3 mmvdn Smfin a: a: a: . a: a: 33
gram mmvdn cumin a: a: a: a: on: «we—
Ewdfi v8.2 «med m3; o3 max” Rb owwd 83
35.3 9.3: m3.w warm m2 emwfi m9 me «new
nnndg emu»: owofi wufim 02 whoa m3 mg 83
2..th onwd v3.3 536 c3 "mg” no“ mmw 83
Enid weed $13 ooNK v3 81m Rb cwwd 2.3
mom—Baum £3332 280
380 85322 bn_~m.eoz wo=nm=w «RED amp—8m a. Eomn— hem
13°F baiw Racism EoEnva .850 mfimaom wimaEm do .26; 30>
.33 S 2.3 .580an .2582
.3 .oz 5er E2 28% €389: 5298 .33 35on 8565.5 Bop 8.25%. “2. uses

 

88
Within the sorghum and millet breeding unit, researchers targeted approximately

seventy percent of their resources to sorghum and thirty percent to millet. Sorghum
research activities included a breeding progam for rainfed sorghum, and varietal
screening trials for both rainfed sorghum and mouskwari. As stated above, the principal
technology generated by these activities was $35, an output of the varietal screening
trials. However, due to the highly correlated and complementary nature of the breeding
and screening activities, the costs of both are included in the sorghum cost stream.

Within the cereal agonomy unit, researchers allocated approximately thirty
percent of their resources to sorghum and seventy percent to maize (Talleyrand, et al.,
1985). Sorghum research within this unit examined the yield stability of “improved"
varieties in a range of cultural and management practices (eg., tied ridging, seed
treatments, fertilizer use, multiple planting dates). Since this research contributed to the
identification of $35 as an appropriate variety for northern, semi-arid Cameroon, its cost
is included in the analysis.

Because actual project expenditures for these activities were not available at the
time of the analysis, operating expenses, administrative overhead costs, and the cost of
the salaries and benefits of the expatriate research staff were estimated (Table 4.16).

Operating expenses were estimated from annual budget requests that had been
submitted by each research unit. Although researchers suggested during interviews that
these budgets were good proxies for actual expenditures, project administrators noted
that in past years budgeted and actual expenditures had varied by as much as plus or
minus fifty percent. Further, budget requests were not instituted within the NCRE
project until 1987, which explains why budget request data were not available for the

period 1982 to 1986. However, both researchers and administrators contend that annual

89
expenditures have not fluctuated significantly over the lifetime of the project and that the

budget requests are the best available proxies.

Hence, for the period 1982 to 1986, the operating cost estimates are based on the
average annual budget requests submitted by the two research units during the four-year
period 1987 to 1990. The averages are adjusted to include only sorghum expenditures by
multiplying the costs of the sorglnum and millet breeding unit by 0.7 and the costs of the
cereal agonomy unit by 0.3. Actual estimates of sorghum’s share of annual operating
costs are 8.4 million fcfa for the breeding unit and 2.7 million fcfa for the agonomy unit.

Administrative costs are estimated as a function of total estimated operating
costs. A twenty-percent administrative overhead charge is assumed, implying that total
annual operating costs are multiplied by 0.2 to derive an estimate for administrative
costs. Hence, estimated annual administrative costs are 2.2 million fcfa.

The salaries and benefits of expatriate researchers are estimated from available
cost data for the NCRE project. An internal document from IITA, the principal
contracto'r for the N CRE project, listed total “professional salaries and allow[ances]" of
$128,960.87 for the month of November, 1985. By assuming that salary and benefits did
not fluctuate significantly on a month to month basis, this total represents the ”average“
monthly costs of salaries and benefits. Given that there were eleven expatriate staff32

involved in the project in November, 1985, the annual cost per researcher is:

(S 128,960.87 / 11 researchers) x 12 months 2: $140,000/researcher/year

 

2“The eleven researchers/ administrators were Almy, Atayi, Dangi, Empig, Everrett,

Janakiram, Kikafunda, McHugh, Roy, Talleyrand and Welsh (NCRE Annual Reports,
1984 and 1986)

90
For the period 1982 to 1986, one expatriate researcher was assigned to each of

the two research units involved in the development of S35. As with operating costs, the
total salaries and benefits of these two researchers are adjusted to reflect the proportion
of total time and effort allocated to sorghum research, multiplying sorghum and millet
breeding salary costs by 0.7 and cereal agonomy costs by 0.3. Hence, the estimated total

salaries and benefits spent on the development of S35 is $140,000 per annum.

Table 4.16 Estimated Total Expenditures (Nominal $US), Sorghum
Research within the National Cereals Research and
Extension Project, Northern Cameroon, 1982 to 1986.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year Operating Administrative Salaries & Total
Expenses Expenses Benefits Expenditures
1982 33,780 6,756 140,000 180,536
1983 29,126 5,825 140,000 174,951
1984 25,400 5,080 140,000 170,481
1985 24,705 4,941 140,000 169,646
1986 32,053 6,411 140,000 178,464
Totals 145,065 29,013 700,000 874,077
Source: NCRE documents and interviews with NCRE researchers
and project administrators.
W

IRA’s financial contribution to the development of S35 is divided into four cost
categories: operating expenses, salaries and benefits of host nationals employed in the
system, administrative costs, and in-kind contributions (eg., buildings, land). Due to data

constraints, the cost streams in this analysis only include estimates for the first three

91

categories, thus underestimating total costs and overestimating the system’s rate of
return. Total costs are modified during sensitivity analysis to estimate the impact of this
underestimation.

For the period 1983 to 1986, total operating expenses for the IRA cereals
progam were provided by the accountant at the IRA-Maroua research center. Since
records for years previous to this were not available, estimates were based on a report
summarizing national IRA expenditures in 1980 and budgeted expenditures in 1981
(Ekelbee). From these records, annual operating expenses for the period 1979 to 1982
were estimated.

The estimates of sorghum’s share of operating expenses for the IRA-Maroua
research center were based on several estimations and assumptions. Expenses for IRA’s
Maroua Research Center are recorded by progams, one of which is the cereals
progam. Within this progam, there were two research units during the 1979 to 1986
period: the sorghum and millet research unit based in Maroua and the cereal agonomy
unit based in Garoua. The first assumption is that these two units received an equal
share of the cereal progam’s budget. Within each of these units, the same allocation of
resources are assumed as is applicable to the NCRE project (70% of costs for the
sorghum and millet unit and 30% of costs for the cereal agonomy unit).

As was the case with cowpea, salaries and benefits of IRA research stafi involved
with sorghum research were not readily available for analysis. Hence, a history of the
staff involved in sorglnum research and estimates of their salaries and benefits was
constructed from IRA and NCRE documents. Included in this history are the research
stafi’ associated with the. SAP GRAD on-farm progam, the sorghum and millet breeding

progam and the cereal agonomy progam, as well as the administrative staff based in

92
Maroua. As before, salary expenditures are adjusted to reflect the fact that sorghum

research represented only a portion of each staff member’s research and/ or
administrative efforts (Table 4.17).
W

As with cowpea, the successful adoption of the irnproved sorghum technology
depended in part on the extension efiorts and input distribution capacity of
SODECOTON. The method for estimating the cost of these activities parallels the one
used with the com analysis. The sole exception is in computing sorghum’s relative
share of total hectares in food crop production. With sorghum, agicultural census data
indicate that, on average, sorghum comprises 73% of food crop hectares harvested in the
Far North Province. Hence, by assumption, 73% of SODECOTON’s food crop .
extension costs are allocated to the sorghum extension cost stream. As with cOWpea, the
estimates were based on SODECOTON’S total annual expenditures for extension stafi'
multiplied by a cost factor. In the sorglnum case, this factor equals (10% "' 73% “ 1.2),
or simply 0.0876. The resulting estimates of annual costs are reported in Table 4.18.
SODECOTON costs are included for the period 1981 to 1986, since IRA, and the
SAF GRAD and NCRE projects collaborated with the parastatal during these years in
both on-farm research and in extending the improved sorghum technologies.
We

Annual costs for the eight-year period during which the sorghum technology was
developed and then extended to farmers are summarized in Table 4.19. The last column,

"Total Annual Costs”, is the cost stream used in the benefit-cost analysis.

93

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.17 Estimated Total Expenditures by IRA, in Nominal ms, for
Sorglnum Research at IRA’s Maroua Research Center,
Cameroon, 1979 to 1986.
Year Research Admin. Operating Total
Staff Staff Expenses Expenses
1979 4,552 3,348 18,642 26,542
1980 4,611 3,505 19,099 27,215
1981 3,977 2,835 14,337 21,149
1982 18,373 2,438 12,068 32,879
1983 18,262 2,186 21,441 41,889
1984 25,503 1,983 18,347 45,833
1985 27,212 2,006 16,409 45,626
1986 36,546 2,706 16,826 56,077
Totals 139,036 21,006 137,168 297,210
Source: NCRE and IRA documents and interviews with the IRA-

Maroua accountant.

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.18 Estimated Total Expenditures, Nominal SUS,
SODECOTON Extension Costs for Sorghum
Technologies, Northern Cameroon, 1981 to 1986.
Year Annual Costs for Annual Costs for
extension stafi, food and Sorghum Extension
cash crops
1981 2,552,541 223,603
1982 2,461,364 215,615
1983 1,839,374 161,129
1984 1,858,072 162,767
1985 2,539,872 222,493
1986 2,507,110 219,623
Total 13,758,333 1,205,230
Source: Estimates based on internal documents and

interviews of SODECOTON staff, 1991.

 

95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.19 Estimated Total Costs, Nominal SUS, Sorghum Research and Extension
Progams, Northern Cameroon, 1979 to 1986.
Year SAF GRAD NCRE IRA SODE- Total
J .P. 31 COT‘ON Annual
Costs
1979 23,881 0 26,542 0 50,423
1980 20,790 0 27,215 0 48,005
1981 18,225 0 21,149 223,603 262,977
1982 18,788 180,536 32,879 215,615 447,818
1983 19,807 174,951 41,889 161,129 397,776
1984 83,799 170,481 45,833 162,767 462,880
1985 91,924 169,646 45,626 222,493 529,689
1986 78,331 178,464 56,077 219,623 532,495
Totals 355,545 874,077 297,210 1,205,230 2,732,063
Source: Tables 4.15 to 4.18.

 

4.3.3 Sorghum Benefit Stream
2 . . l H E g

The benefit stream associated with sorghum research in northern Cameroon was
estimated from data on: (1) farmers’ yields for local sorghum varieties and for $35, the
improved variety extended to farmers; (2) the frequency of drought conditions; (3)
annual adoption rates of the improved technology; (4) hectares in sorghum production;
and, (5) market prices for inputs and outputs.

These data are the basis for the estimates of the market value of the annual
aggegate gain in sorghum production due to the development and extension of S35.
Thus, this value, in US dollars, quantifies the gains from sorghum research and forms the
benefit stream used in the benefit-cost analysis. 5
5 l E . 1:. l !

Numerous sources report sorglnum yields for one or more sets of farming
practices, including trial data reported by the SAP GRAD and NCRE projects, and on-
farm yields reported by SODECOTON and MINAGRI. Crop management practices
include both intercropped and monocropped systems, although monocropped systems
vary considerably in the degee and breadth of adoption / application of improved
technologies. Reported yields (Table 4.8) range from 448 kg/ha for a TLU-sponsored
on-farm intercropped sorghum trial to 5,588 kg/ ha for a CRSP-sponsored on-station
cowpea/sorghum intercropped trial (NCRE Annual Report, 1989; Ta’Ama, 1988,
respectively).

Sorghum yields are estimated by combining available yield data with qualitative
data on rainfall patterns in northern Cameroon. Although the data in Table 4.8 indicate

yield potentials, yields in any given year are highly dependent on the quantity, timing and

97
dispersion of rainfall in the region. Further, sorghum researchers concede that the

improved variety outyields local varieties only in years when the onset of the rainy
season is late and / or total rainfall is below average.

Hence, there was a need to estimate three yields: an average yield for both
traditional and improved varieties under normal rainfall conditions, an average yield for
the improved variety under drouglnt, and an average yield for traditional varieties under
drouglnt. An estimate of the probability that rains will be late and/ or below normal
levels was also needed.

Estimated yields for farmers producing sorglnum under normal rainfall conditions
are the most similar to published results. Although Agicultural Census data and
SODECOTON estimates set farmers’ yields relatively higln, ranging from 650 to 1,467
kg/ha, one sorghum researcher in IRA suggested that sorghum farmers in the Center-
North zone have yields averaging 600 to 700 kg/ha. While trial results surpass either of
these yields, some by almost a factor of 10, they were discounted since on-station trial
results have limited predictive power of farmers’ yields. Hence, sorghum yields for
normal rainfall conditions are set at 800 kg/ ha. This value represents the middle ground
of available data”.

Yield estimates under drought conditions are based on trial data from 1984, a
drought year, and from anecdotal evidence provided by key informants. Yield data from
on-farm varietal screening trials indicate that S35 yields in 1984 averaged 73% of yield

for S35 during the subsequent three years (IRA/ SAF GRAD /TLU, various reports).

 

3"Sorghum yield was estimated to be 600 kg/ha for the cowpea benefit-cost run. This
discrepancy is intentional. Intercropped cowpea and sorghum are generally gown in
more droughty areas of the region, implying lower average yields for intercropped
sorghum relative to average region-wide yields.

98
Similarly, traditional varieties yielded only 42% of the subsequent three year average.

Even more dramatic gains for S35 were suggested by an IRA researcher who proposed
that in drought years, S35 will typically outyield local varieties by 500 kg/ha. Based on
this evidence, drought-year yields were set at 650 kg/ha for S35 and 300 kg/ha for local
varieties.

For northern Cameroon, limited time-series data on rainfall are available. IRA-
Maroua reports annual rainfall data par décade (by ten-day periods) for twenty sites in
the cotton gowing area of its research zone. These series, depending on the site, extend
back in time for fourteen to thirty-six years. Eleven of these sites are in the same area
where the highest levels of S35 adoption have been reported.

As a crude estimate of the probability of drouglnt at these sites, the ratio of years
with total rainfall below 600 mm to total years in the times series was calculated for each
site. The average value of this ratio was one drought year in every four years. However,
this ratio only considers total rainfall and not the possibility of late onsets of the rainy
season. Also, as noted in Chapter 2, Section 2.2, northern Cameroon has experienced an
extended drought period for the last fifteen years, and key informants in northern
Cameroon speculate that drought conditions may occur at any given site in any given
year. Hence, the benefit-cost analysis sets the probability of drought at one in every
three years, implying triennial yield patterns of 800, 800, 650 kg/ ha for S35 and 800, 800,
300 kg/ ha for local varieties.

Welds

Although in northern Cameroon farmers primarily gow sorghum for gain, stover

is an important and valued by-product. Stalks are used as livestock feed during the dry

season and as building materials for fences, roofs and walls. S35 is a medium height (2.5

99
meters) variety, a trait cited by some farmers as a reason for nonadoption (Kamuanga,
1990). Clearly, these farmers value the longer stalks common to many of the local
varieties.

By adopting S35, farmers forego the preferred stalk length and/ or total quantity
of stover production, traits that have some econonnic value. However, no data exist
indicating the net loss in stover production from adopting S35 nor its economic value. In
the cowpea section of this thesis, an estimate of stover production and its market value
were reported. Although similar estimates could be made for this analysis, key
additional information needed to estimate the stover losses are still absentuthe net
difierence in stover production between improved "medium height" and traditional ”tall”
varieties, and the percentage mix of tall, medium and short varieties traditionally gown
by farmers. Without these data, it is impoaible to estimate of the loss of stover
production. Hence, this analysis ignores any loss in stover production, and consequently,
overestimates the returns to sorghum research (although this overestimation is probably
quite small).

W

A prefatory statement is necessary to understand the potential level of adoption
of S35 in northern Cameroon. It is provided by Kamuanga, who notes that:

North Cameroon is a region endowed with an enormous variety of local

materials. More than 1,800 accessions have been evaluated by IRA in

collaboration with ICRISAT (Dangi et al., 1989)...Any new introduced variety in

the release process should be considered as a complementary addition to the
farmers’ ‘pool’ of varieties and not as a replacement of local materials (1991,

p. 12).
Given 835’s advantage in drought conditions, its adoption should depend, in part,
on the climatic conditions fawd by farmers. If farmers could perfectly predict drought

years at planting time, they would choose the appropriate variety for the expected levels

100
and timing of rainfall. Clearly, this is not possible. Only in years when the onset of the

rainy season is late can farmers plant S35 with the explicit purpose of capitalizing on its
drought tolerance and short gowing cycle.

Yet, farmers have adopted a strategy that incorporates the improved variety as
an apparent attempt at lowering risk. Some adopters of S35 crop the improved variety
with traditional varieties in the same field, thus diversifying the crop mix in the field and
lowering their risk exposure. Although Russell (1991) (found few advantages to these
mixed sorghum stands in northern Cameroon, he notes that, “farmers themselves are
gowing mixed stands, even with the newly-introduced varieties, [which] indicates that
more information is needed on farmers’ strategies and motives (p. 158).“

Since rainfall patterns in northern Cameroon are so erratic, S35 may be the
appropriate variety for any given site in any given year. As Kamuanga notes, ”the
diversity of sorghum gown in north Cameroon makes it possible for farmers to select,
for any given planting date, those varieties that are very likely to mature with or soon
after the rains (1991, p. 9).” Hence, the adoption of 835 on a per hectare basis was
assumed to be relatively constant from year to year, although the same farmer may not
gow S35 in two consecutive years.

Annual adoption rates are quantified using available indicators of adoption,
including production figures from IRA and SODECOTON documents, seed sales
reported by the NCSM project, and an adoption survey conducted in 1990 by the TLU.
These data are incorporated into a log'stic function to predict annual adoption rates and
the adoption ceiling.

Initial adoption figures (years 1984 and 1985) were limited to on-farm trials

conducted by the SAP GRAD J.P. 31 in collaboration with IRA and SODECOTON.

101
Technically, these figures do not represent actual adoption since collaborating farmers

were simply following a strict protocol, including which varieties to gow. However,
these trials do represent a form of extension whereby improved varieties were introduwd
to local farmers.

In 1984, 7 ha of S35 were gown in farmers’ fields as part of the on-farm research
progam; similarly, 24 ha were gown in 1985. These areas seeded to S35 represent
0.004% and 0.012%, respectively, of total area in sorghum production. Given that 1984
was a drought year in which S35 clearly and consistently outperformed all other available
varieties, it is highly likely that many of the farmers involved in on-farm trials in 1984
continued to gow S35 in 1985. However, there is no record of this level of adoption,
implying that the 1985 adoption estimate of 0.012% is an underestimation, perhaps by as
much as a factor of 10. The significance of this underestimation is tested during
sensitivity analysis.

In 1987, Jerry Johnson surveyed farmers who had participated in on-farm trials
during one of the previous three gowing seasons (1984 to 1986). He reported that 23%,
29% and 48% of farmers who participated in 1984, 1985, and 1986, respectively, still
gew S35 in 1987, indicating that some farmers had adopted 835. This is confirmed by
SODECOTON production records which indicate that in 1986, 649 ha of S35 were _
harvested as part of the parastatal’s food crop extension progam. But SODECOTON
data underestimate the adoption of S35 in 1986, since in that year the NCSM project
produced and sold 42.34 metric tons of S35 seed At the recommended planting density
of 20 kg/ha, this represents 2,117 ha seeded to S35. Since 1986 was the first year that
S35 was extended to farmers on a large scale, seed sales are assumed to be the most

accurate proxy for total adoption for that year. This implies an adoption rate of 1.03%.

102
Seed sales in subsequent years were not as good a proxy for adoption since the

improved variety is open pollinated, implying farmers can propagate their own seed
stock. Hence, in 1990, motivated by the need for a more accurate measure of adoption,
the TLU conducted a large-scale adoption survey throughout the Center-North zone,
focusing primarily on the adoption of S35. Results from this survey indicated that S35
comprised 3.3% (pure stand equivalents) of the sorghum area harvested in 1991
(Kamuanga, 1991).

Using the adoption rates of S35 in 1986 and in 1990, the parameters for the
logistic function (Equation 4.3) were calculated by solving the following two-variable,

two-equation algebraic problem:

1.03 + (1.03 t be") = K (4.4)
3.30 + (3.30 It be") 8K

The parameter values for K and beta are 3.44 and 47, respectively, where K is the
adoption ceiling, in percentage of land planted to sorghum.

Using equation 4.3 and these parameter values, annual adoption rates were
calculated for years 3 to 15 of the benefit stream (1986 to 1998). Since actual adoption

figures, in the form of on-farm trial results, are available for the first two years that S35

103 .
was extended to farmers, these are used for 1985 and 1986. Annual adoption rates and

hectares in production“ are reported in Table 4.20.
W

Sorglnum production in northern Cameroon requires, at a minimum, seed, labor
and land. Extension recommendations add seed treatment and fertilizers to the list of
possible inputs, which, if used, clearly add to the cost of production. However, for this
analysis, the key point is not the actual cost of inputs but whether expenditures on inputs
at the farm level changed with the adoption the improved variety.

As noted in the sorghum yield section above, researchers concede that $35 does
not outyield local varieties in normal rainfall years. Implicit in this comparison is that all
other levels of inputs are the same. The analysis requires that one of two simplifying
assumptions be chosen: either assume that the level of inputs used by an individual
farmer is independent of the variety gown or that farmers gowing S35 achieve higher
yields during normal rainfall years because the adoption of S35 implies a more intensive
level of input usage. Anecdotal evidence suggests both assumptions are true for some
farmers, although the former appears to be the dominate behavior. Thus, this analysis
assumes that farmers choose levels of inputs independent of the variety(ies) grown, so no
additional farm-level input costs are associated with the adoption of S35. Consequently,

farm-level input costs were not included in the benefit-cost analysis.

 

”Total area harvested for rainy season sorghum ("first crop" sorghum) is estimated
from Agicultural Census data for the Far North Province plus the Mayo Louti
Department of the North Province. These estimates are detailed in Section 4.3 .4 under
the heading "Area Harvested”.

104

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.20 Projected Adoption Rates and Estimated Area
Harvested for S35 Sorghum, Northern Cameroon,
Years 1984-1998.
Year Total Area S35 Adoption Total Area
Harvested, Rainy Rate, % of Harvested, S35
Season Sorghum area harvested
(Ha) (Ha)
1984 172,260 0.004 7
1985 208,577 0.012 25
1986 205,607 1.030 2,117
1987 114,787 1.849 2,122
1988 183,700 2.613 4,799
1989 193,516 3.081 5,962
1990 179,800 3.299 5,931
1991 179,800 3.387 6,089
1992 179,800 3.420 6,149
1993 179,800 3.433 6,172
1994 179,800 3.437 6,180
1995 179,800 3.439 A 6,183
1996 179,800 3.440 6,184
1997 179,800 3.440 6,185
1998 179,800 3.440 6,185

 

 

 

 

 

105
W

Quantifying the economic value of sorghum production in northern Cameroon for
improved varieties requires time-series price data for white sorghum gain. Twa average
prices are needed for the analysis: the price of white sorghum during normal rainfall
years and the price during drought years. Kamuanga, et al., report that sorghum grain
price fluctuations are extreme, varying by as much as 400% in response to supply
fluctuations.

MINAGRI reports monthly retail prices for white sorghum over a six-year period
(1985 to 1990) for five "urban“ markets, as detailed above in Section 4.2.3. Since this
series is incomplete, prices in representative years are used as proxies for average annual
prices.

Since the 1984 gowing season was a drought year, pre-harvest prices in 1985
reflect the economic value of sorghum production after drought years. Prices, available
for three markets (Maroua, Kaélé and Mora) for the months of July through September,
ranged from 160 to 206 fcfa/kg of white sorghum. Given the severity of the drought in
1984, these prices probably represent an upper limit of the normal range of sorghum
prices. They also are 'hungy season” prices, representing not the price at harvest but
rather the price eight to ten months after harvest. Hence, for the base run of the
analysis the drought year price for sorghum was set at 130 fcfa / kg, approximately twice
the normal rainfall year price.

The normal rainfall year price is based on the average market price for white
sorghum across all five markets for November, 1988—which ranged from 35 to 75 fcfa/ kg

and averaged 63 fcfa / kg. November prices were chosen since this month represents the

106
peak period of sorghum harvesting and marketing by farmers. The year 1988 was chosen

since rainfall that year was at a historically normal level for the region”.

These two prices are incorporated into the benefit stream as part of the
assumption that drought years come once every three years. Thus, a triennial price
pattern of 130, 63, and 63 fcfa/kg is repeated throughout the time-frame of the analysis,
beginning with 1984 as a drought year.

4.3.4 Base Run for Sorglnum ROR
mama

Two estimates are needed concerning area harvested: the annual total harvested
area and the annual area harvested for the variety S35. The former is an estimate based
on Agicultural Census data. The latter is simply the former multiplied by the adoption
rates reported in Table 4.20.

To use Agicultural Census data to estimate area planted required two sets of
adjustments. First, annual data on area harvested are only disaggegated to the
provincial level. However, the extension zone for 835 includes the northern most
department (Mayo Louti) of the North Province in addition to all of the Far North
Province. Since disaggegated, department-level data were only available for 1984, the
ratio of sorglnum area harvested in the Mayo Louti Department to the total sorghum
area in the North Province is assumed to be constant over time. Multiplying this ratio
times the annual provincial-level data of the North Province gives an estimate of the

sorghum area harvested annually in the Mayo Louti department. Second, as noted in

 

3sThe Maroua TLU reported monthly prices for six rural markets which were
averages of 1989 and 1990 prices. The overall 24 month average price was 64 fcfa/ kg,
while average price in November was 56 fcfa. These data combine white and red
sorghum prices, even though white sorghum commands a premium in the market place.

107
Chapter 2, the Agicultural Census reports sorglnum and pearl millet data combined

together. Hence, census data are multiplied by 0.9 to remove millet from the estimated
totals.
W

Total production is estimated by multiplying yield estimates by the areas
harvested. Annual production totals for the two cropping systems, given the two rainfall
scenarios, are then aggegated to determine the total annual production, given the
introduction of S35. A second total production figure is estimated for the region, based
on the assumption that the improved package had never been developed and extended.
The difference between these two totals represents the gain in production due to the
adoption of $35. The market value of this gain is then estimated, using the price data
and exchange rates discussed above.
W

Gross benefits from the development and extension of 835 are simply the annual
market values of the gains in production, converted to SUS. The estimate of goss
benefits is based on several key assumptions, summarized as follows: (1) stover loss
from the adoption of S35 is marginal and therefore can be ignored; (2) farmers choose
their level of inputs (seed treatments, fertilizer, etc.) independent of the variety gown;
and, (3) rainfall has a triennial pattern where "good” rains occur two out of three years
and drought occurs during the third.

The time horizon of the benefit stream is fifteen years, beginning in 1984, the
first year $35 was tested widely in farmers’ fields. Key informants generally believe that
$35 is now "out there" as part of the pool of sorghum varieties from which farmers select

each year. Because the variety has been extended widely and has noticeable advantages

108

during drought conditions, the assumption that its benefit stream will continue for
another seven years from the time of this analysis (1991) is relatively conservative.
W

The internal rate of return for the base run is 0.9%, calculated for the net
cost /benefit flow reported below (Table 4.21).

4.3.5 Sensitivity Analysis - Sorghum

Although the base run estimate of the IR is the best judgnent possible of the
returns to sorghum research and extension in northern Cameroon, sensitivity analysis is
conducted to test the robustness of the estimate. Key assumptions and parameter values
are tested to see how changes in their values aflect the reported IRR.

Approximately forty alternative sets of assumptions and / or parameter values
were tested, and IRRs were calculated for each of them. Ten of these runs, judged to be
the most telling, are reported in Table 4.22 and are discussed below. With these runs,
IRR values ranged from -5.6% to +7.9%.

The reported IRRs generally differ only slightly from the base run, indicating that
the IRR estimate is relatively robust. The values of the IRRs tend towards zero to
slightly positive, indicating that sorglnum research and extension probably broke even
(i.e., was "able to pay for itself” in financial terms), but most likely failed to earn
suflicient returns to be “profitable” in economic terms (i.e., where “profitability " implies
returns geatcr than 10%, the approximate opportunity cost of capital in northern

Cameroon).

109

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.21 Estimated Benefit-Cost Flows (ill ’000 $US) for the Development and
Extension of the Improved Sorghum Variety S35, Northern Cameroon,
1979 to 1998.
Year Gross Benefits Gross Costs of Net
of S35 Research & Benefit
Extension flow
1979 0 -51 -51
1980 0 -48 -48
1981 0 -263 263
1982 0 -448 -448
1983 0 -398 -398
1984 1 -463 -462
1985 0 -530 -530
1986 0 -530 -530
1987 294 0 294
1988 ’0 o ' o
1989 0 0
1990 601 0 601
1991 0 0 0
1992 0 0 0
1993 943 0 943
1994 0 0 0
1995 0 0 0
1996 1,119 0 1,119
1997 0 0 0
1998 0 0 0

 

 

 

 

 

1m

devisgggfiaefiofizsganiggr

devsfibavoteneguegflooiee-ofiucsgaasgg

sSQEiBfifiigasiiiifsgfiaiiafie§¥§§a§§€aflfliifii§§

a§§£§f§§§§8§§§Qr

 

 

33588358835
3333835338.

aiitsiasfififi
2358833338.
ue§Z§

a88983§

 

 

a88888§§m

 

A835 835.80 .5995 .3 begga—

 

3332532

 

595855.5-
A3555£Eifisfia

 

§§§3%€§§§§

 

nfii§§5§i§
§$65§3§i§§§

 

83Efi§§5§§§

 

“aaaaaaaa

t
a

3

aifiggflgfifi
Aggrifiigaafigr

 

 

 

 

 

 

i

 

 

lfiéié

 

d§éa§§§9§€£§§§fl
€332§i§§55§££2503%58B§§§SE§3§§8§3¥§€2£§§¢

«N6 039—.

 

111

3.385.95533355i5353882... 33883....
.85 .855. a... 2.6.5. .558: 3833...... £38.. 32.....8 5.. 8.2.8. o... 8.1.... a... 8.2.8. o... .2. 8a....- .. .. .5: a... .8...

5.. an..." a 83...... 3.8:... .2- 3....- o-epi 2.. 5...: .6. at?

o... a. 8.5.2 2. a... 292.: «3...... £28 8.22.... .5... «as... 88...». 3.8:... 8.. 3.5.8 .35... 8. a... 29.2.: a... s... 8..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

«6+ 9.5+ ta. 03...! .o 0...! c. 835... =-
Wo. n..- Wm... 03...... 3 039 E 33.3.. a
n.~. :9. o.n+ .52.. My: .0 o.._u> E 993.5
0... o... WT 03:19 .0 2:9 5 83.6... =-
06. to. «.v 035...... .0 02!. a. 8.28.. a
v..- w... an a... no. 3... 5..
.8. u a a3 .. a .. fin u .. ..§.~ a 8.... :80 .so...
a «8.. .. a a .. u8a.. a 8.... 280 8.22...
35.96:...

a a a is»... a a 3:5... .335 .8228 .585 .c .0598...
.8. u a a a a he...” a $8.2. .2. 8.8.5: 82
R... n a a a a a a 8. 55 5.3.8

"9.3... 23> 29.2.. a. 8....

.. a 8.. a a a 8.. 3.8...» .82 a...» 62.99

a 5 8a a a a 83 gm but! .5...» E393

".13. 59> .352 a. .2...»
.2. u a a a a .. .. 8a 8.3.5. :8. .53 5399
*3 u n a 2:. a a a one a but! .53 £2.93
as}... an; ES... 5 6.»;
= .5... o. .3... .a .3. a .5. n :3. o E... .5. a... 33...; 3.
$8.58. .8236 .5582

.833... q .383. 9.5.8 .6. :9. a... a. 89.20 33533. .2... 83...; 3. .o 82.5 9.5.3... «1.32. £3.18...

J63 .NN‘ 030.".

 

112

W

The drought year yield for the improved variety was varied by plus or minus 25%
(2 162.5 kg/ha), resulting in IRRs of 5.2 and -5.6%, respectively. The increase in the
IRR is expected with an increase in the yield of S35. Conversely, the IRR declines with
a decrease in 835’s competitive advantage (i.e., lower yields) over local varieties in
drought years. The higher IRR is somewhat suspect since a 25% increase in drought
yield implies that S35 has a slight yield increase in drought years when compared to its
yield in normal years (812 verses 800 kg/ha). Clearly, this is unlikely and the analysis
was adjusted accordingly, setting both 835 yield parameters at 800 kg/ha, resulting in an
IRR of 4.9%. Conversely, a 25% drop in the value of 835’s yield in drought years
drastically affects the IRR, lowering it to -5.6%. A more conservative 10% drop inithis
parameter’s value resulted in an IRR of -1.3%, implying that the analysis is sensitive to
the assumptions made about 835’s drought yield.
WWW

The IR was less sensitive to changes in assumptions about the drought yield for
local varieties, relative to similar changes in the drought yield for 835. With a 25 % plus
or minus change in the parameter’s value (:75 kg/ha), the IRR changed to -1.7 and
3.0%, respectively. A decrease in the returns to research and extension is expected if the
”defender crop“ is assumed to be more competitive. Such is the case when the drought
year yield parameter for local varieties is increased to 375 kg/ha. The opposite is true
as welluif the “defender crop" becomes less competitive, the benefits of the improved
variety increase. The likelihood of yields falling within the 225 to 375 kg/ ha range
depends almost entirely on the severity of drought assumed and / or experienced. Hence,

interpretations of this run are difficult, and limited to an obvious conclusion: as drought

1 13
conditions become more severe and / or frequent, the more likely the benefits of $35 will

translate into positive returns for research and extension.
W

The returns to research and extension were also sensitive to the assumptions
made about sorghum prices in drought years. Again, this parameter’s value depends on
the severity of the drought and the magnitude of the resulting shortfall in supply. A 25 %
decrease in the drought year price resulted in a decrease of over three percentage points
in the IRR estimate—falling from 0.9% to -2.2%. An increase of 25% in drought year
prices led to an IRR of 3.3%. Given the extreme volatility of the sorghum market in
northern Cameroon, a drought year price range of 98 to 162 fcfa/ kg is quite possible.
However, anecdotal evidence suggests that to set prices in drought years are even higher
than those proposed in the sensitivity analysis, implying that the price parameter in the
base run IRR is more likely to underestimate rather than overestimate the IRR.
W

Total sorghum area harvested is varied by plus or minus 25 %, resulting in IRRs
of -2.2% if the area is reduced and +3.4% if the area is increased. Identical results are
obtained when varying the adoption rate by plus or minus 25%. These results indicate
that as more land is cropped to S35, the returns to research and extension increase.
They also point out that the results of the analysis are sensitive to area estimates drawn
from the Agicultural Census.
W

Aggregate costs of the sorghum research and extension program are reported in
Table 4.19. One noteworthy statistic from that data is that extension costs for the

sorghum program represents approximately 44% of total estimated aggregate costs.

114

Although this analysis assumes that part of the success of 835 research, in terms of
farmer adoption, is due to the «tension efforts of SODECOTON, an alternative
assumption is to ignore «tension efforts and only calculate an IRR for sorghum
research. With no «tension costs included in the cost stream, the IRR for sorghum
research is 7.7%. Given the high degree of collaboration between IRA and
SODECOTON, and the breadth of SODECOTON’s «tension program, «eluding
«tension costs is suspect because it likely ignores «penditures that were critical to the
adoption of the improved variety. However, due to methodological inconsistencies in the
literature as to what costs to include or ignore, this alternative is presented.
W

Estimates of total annual costs were varied by plus or minus 25%, resulting in
IRRs of -1.5% and 4.1%, respectively. Cost data for the sorghum program were
incomplete, requiring the use of proxies and informed assumptions. Only in the case of
the SAF GRAD project were cost data readily available. Hence, the IRR may be under-
or overestimated due to errors in estimating total costs. The sensitivity analysis indicates
that the IRR estimate is sensitive to total costs, although it is more sensitive to
overestimation.
W

The two data bases concerning total sorghum area harvested are discussed in
Chapter 2, as are the justifications for using DEAPA’s data over MINAGRI’s. However,
to test the significance of choosing one data base over another, MINAGRI’s data were
substituted into the analysis, resulting in an IRR of 3.9%. The analysis assumes that the-
total amount of 835 harvested is a proportion of total sorghum area harvested for all

varieties. Since the MINAGRI estimates of total sorghum area harvested are

115
consistently higher than estimates made by DEAPA, the use of MINAGRI data leads to

a higher adoption rate (in terms of area harvested) and thus to a higher IRR estimate.
W

A very critical assumption, from a theoretical point of view, pertains to the
frequency of drought conditions in northern Cameroon and its subsequent impact on
overall sorghum production. Given the assumption that the improved varieties only
generate benefits in drought years, the hypothesized frequenq' of drought conditions are
fundamental to the IRR estimation. The base run estimate fixed a three-year rainfall
pattern of two years of normal rains followed by one year of drought. Sensitivity analysis
was used to test the significance of this estimated pattern of rainfall. Two alternatives
were tested-drought conditions once every four years and drought conditions once every
two years. The former resulted in an IRR that was just slightly negative, 04% while the
latter resulted in an estimated IRR of 7.9%, the highest IRR reported in Table 4.22.
These results highlight the advantage of $35 in drought conditions and the potential for
payoffs to research targeted to marginal production conditions.
BMW

This run tests the assumption that input use is independent of the adoption of
the improved variety. An alternative assumption is that farmers adopt an entire package
of inputs, including the improved variety, seed treatment, specific management practices
(eg., planting in lines), and fertilizer. This assumption implies that 835 will yield more
than local varieties, even in normal rainfall conditions due to higher input usage.
However, these inputs increase the cost of production, a change which must be factored
into the analysis. With yields and costs adjusted accordingly, the resulting IRR is 0.8%,

virtually the same as the base run estimate. Run 9 increases the confidence in the base

116

run IRR estimate, given that a very difi'erent, although plausible, set of assumptions led
to essentially the same conclusions-sorghum research, in financial accounting terms,
*broke even”.
WW

Given some anecdotal evidence that Cameroon’s currency is overvalued, this run
estimates a shadow «change rate and tests the assumption that inputs and outputs
should be valued at the market «change rate. The methodology is identical to that used
with the cowpea sensitivity analysis (Run 13: Shadow Exchange Rate). The resulting
IRR for sorghum is -2.3%. It is lower than the base run IRR because the value of the
outputs, when converted to SUS, is less after devaluation.
W

A best case/worse case scenario is tested in Run 10, whereby five of the key
variables were simultaneously modified by plus or minus 10%. The resulting IRRs were
7.0% and -5.6%, respectively. Although the worse case scenario implies a negative rate
of return to sorghum research and «tension, the general trend in the sensitivity analysis
has indicated a program which had a zero to slightly positive IRR. Either «treme (best
or worst) is unlikely since it is improbable that all five parameters were misspecified in
such a way that they all affect the IRR estimate in the same way. This run simply
confirms that sorghum research and «tension in northern Cameroon was not a
resounding economic success. However, sorghum research throughout West-Central
Africa has been notoriously difficult (Johnson, 1987; Kamuanga, 1991), implying that a
program that at least ”paid its own way” in financial accounting terms is an exception and

a relative success.

CHAPTER 5 INSTITUTIONAL ANALYSIS

The general objective of this chapter is to analyze the institutional factors,
linkages and characteristics associated with the research-«tension system of northern
Cameroon. This analysis will help determine how these institutional traits interacted to
complement and/ or impede the performance of the cowpea and sorghum subsectors.
The chapter is divided into. three sections, the first being a detailed, forty-year
chronology of agrimltural development activities in northern Cameroon. The latter two
sections focus on a specific time period, 1979 to 1986—the period during which the
improved technologies discussed in Chapter 4 were developed. This two—part section
first evaluates linkages within the research-«tension system and assesses how these
linkages affected the system’s performance. This is followed by a brief discussion on the
distribution of benefits resulting from the development and «tension of the improved
technologies.

5 .1 Agricultural Development in Northern Cameroon

Agricultural development efforts in northern Cameroon have spanned forty-years.
The following discussion divides this history into four distinct. periods, each of which is
identified by a particular development theme and / or project. Although the following is
an incomplete reckoning of the past“, it highlights the defining events and key
institutions that guided northern Cameroon’s agricultural development and

transformation from 1948 to the present (1991).

 

3“For a more complete history, refer to Bilan Diagnostic du Secteur Agficole de 1960 d
1980, published by Cameroon’s Ministry of Agriculture March 1980.

117

118
5.1.1 Efforts Prior to 1970s - King Cotton

Records indicate that agricultural research in northern Cameroon began in 1948
with the establishment of an agricultural research station in Guetalé, located northwest
of Maroua on the Koza-Mora road.” The limited documentation found describing this
early period suggests that research focused on collecting basic data (eg., the identification
of rainfall patterns, soil types, farming systems). A French firm, la Campagne Frungaise
pour le Déwloppcment dc: Fibres Textiles (CFDT), introduced cotton to the area three
years later, although CFDT had been promoting cotton production in neighboring Chad
prior to its 1951 arrival in northern Cameroon. In 1952, a French research organization,
I’Irmitut de Recherche sur le Coton ct Fibres Textiles (IRCT), established a cotton research
station in 'Maroua. With CFDT and mar, the French laid the foundation for the cotton
industry that still exists today.

Food crop research began in 1964 when the French agency, l’Institut de
Recherche: Agronomiques Tropicales ct den Cultures Vivriéres (IRAT), established a food
crop research station in Guetalé, which they relocated to Maroua in 1972. Although
IRAT research targeted the principal food crops of the region (sorghum, maize,
goundnut and cowpea), its implicit goal was to enhance cotton production.

Cotton production has become a way of life for two generations of Cameroonian
farmers. But it was during this establishment of this cash crop system that agiculture in
northern Cameroon was significantly transformed. Numerous improved production
techniques (i.e., chemical fertilizers, animal draught power, row planting and seed

treatment), as well as more general progams to promote farmer literacy and to improve

 

3"This statement implies a very narrow definition or what comprises agricultural
research, and is limited to research based on “the scientific method”. Local farmers have
been researching agriculture in a less formal sense for centuries, if not millennia.

119

household nutrition were introduced in the name of cotton. Subsidized by the French
and Cameroonian governments, these activities had a dramatic impact on the farming
practices of northern Cameroon.
5.1.2 Early 1970s - Era of Government Interventionism

As the newly founded nation of Cameroon came into its own, government began
to «pand its role in promoting development. As part of this gowing sense of a national
purpose, the government recognized the critical importance of agiculture and adopted
numerous policies to redress traditional sources of uncertainty and production
constraints: limited market outlets for farm produce, unavailable credit, unstable prices,
and an underdeveloped rural sector. In addition, steps were taken to nationalize various
components of the agricultural sector, including research, and specific subsectors, namely
cotton and rice. The following describes several key institutions created during this
period of government interventionism-othree government agencies, two large-scale rural
development projects, and one agicultural parastatal.
W

In 1973, MIDEVIV and FONADER were created to address two fundamental
market failures—unavailable credit for farmers and inaccessible consumer markets.
MIDEVIV, an acronym for Ia Mission de Développement des Cultures Vivriém,
Marutchéres ct Fruitiercs, was mandated to improve village-level market access to food
crops. In practice, the agency was to purchase food crops in areas where there were
surpluses and then transport these commodities to food deficit areas (urban centers
and/ or rural areas suffering from production shortfalls). F ONADER, an acronym for
Fonds National de Développement Rural, was mandated to supply short and intermediate

term loans to individual farmers. When the first OPEC oil crisis occurred in 1973,

120
FONADER was mandated to supply credit to enable farmers to purchase fertilizer

whose price had increased considerably due to the oil shock. Even though the oil shock
provided an important impetus to the formation of FONADER, it was the recognition of
the general need for agicultural credit that led to the agency’s creation and
continuation.

Since FONADER declared bankruptcy on June 30, 1988, only MIDEVIV is still
functioning (1991). Neither agency appears to have been particularly successful, as both
were constrained by limited capital resources and inadequate government support.
MIDEVIV faced the additional problem that surplus agicultural products tended to only
flow one direction. Althougln surplus fruits and vegetables from southern Cameroon
could be marketed in the northern region, there was little demand in the south for the
north’s surplus gains-sorghum and millet had little appeal to consumers, and Asian rice
imported into Douala was less «pensive than locally produced rice from SEMRY. Key
informants observed that the agency simply suffered from too many defaulted loans, poor
management, and corruption.

W

Office Céréalier, founded in 1975, was created to address the problems of price
instability in the cereal markets of northern Cameroon. As noted earlier, supply and
prices in northern Cameroon are prone to «treme fluctuations within a given marketing
year, as well as from year-to-year. To address this problem, Office Céréalier was
mandated to buy gains (millet, sorghum, maize, and rice) when market supply was high
and store these gains for resale at subsidized prices when market supply became low.

In carrying out these activities, the agency had an additional mandate to address food

security issues and maintain emergency food stocks.

121
Internal documents summarizing Office Céréalier’s purchases for the eleven year

period 1979/ 80 to 1989/ 90 indicate that the agency’s annual purchases never «ceeded
5,000 metric tons of sorghum and millet, or 4,000 metric tons of maize. For either
commodity, total purchases rarely represented more than 10% (usually much closer to 1
to 3%) of estimated total production for the two northern provinces. With such low
levels of market intervention, the agency had minimal influence on market prices.
Granted, on days that the agency sold or bought, it is likely that in the very short term
(same day, perhaps same week) local spot market prices would be affected. However,
regional price levels were probably little influenced by the agency’s actions.
SEMRX

Northern Cameroon’s first large scale development project was In Société
dWon et de Modemisation de la Rizicultwe d Yagoua (SEMRY), an irrigated rice
project on the Logone River that involved a huge capital investment (eg., the creation of
a 300 km2 lake) and drastically transformed the region’s ecology and farming systems.
Initiated in 1971, the SEMRY project is currently funded by the BBC (1991), although
its first two phases were funded through the FAC. Probably the most telling
commentary on the success of the initiative is that although SEMRY’s rice production
costs are among the lowest in West Africa, it is still cheaper to import rice into Douala
and transport it north”. Yet, since food self-sufficiency has been an important policy
consideration for the Cameroonian government, SEMRY was touted as a demonstration

of the production capability of the country’s northern region.

 

3"This comment is based on conversations that the author had during the two and a
half years that he lived and work in the region as a Peace Corps volunteer and on
comments made by key informants during interviews conducted for this study.

HEB

In 1972, a second large-scale development project was initiated in the upper
valley region of the Benoue River, east of Garoua. This integ'ated rural development
project, Projet Nora-Est Benoue (NEB), is in its fourth phase (1988-1992) and has
historically targeted a wide range of issues including infrastructure development,
resettlement of farmers, public health and education, as well as agricultural production.
NEB has had a marginal role in the transformation of the cowpea and sorghum
subsectors of the Far North Province, mainly because the project’s «tension zone
targeted a region outside of the province. However, most of the farmers participating in
NEB’s resettlement progam migrated from the densely populated Mandara Mountain’s
of the Far North Province, a pivotal area for W and sorghum production. 5
Quantifying the degee of rural migation and its influences on agicultural production
are beyond the scope of this analysis. NEB is mentioned principally to document one of
the earliest government efiorts to initiate a comprehensive, integated rural development
project in the northern region of the country.

52122991121!

In 1974, CFDT was nationalized, creating the Cameroonian parastatal
SODECOTON. In the same year, CFDT’s sister organization, IRCT, was placed under
the newly formed Office National de la Recherche Sciennfique et Technique (ONAREST),
a national umbrella agency overseeing all facets of agicultural research in Cameroon.
The relative "success“ of SODECOTON and the agicultural research system are
discussed in Chapter 4 of this thesis, but are mentioned here as additional examples of

the nationalization policies pursued in the early 19703.

123
5.13 Early 1980s - The PCN Era

By the early 1980s, the agicultural research and «tension system had reached a
critical mass in its development. All of the pieces were in place: a functioning
agricultural research system, a large, active «tension system, a capacity for seed
multiplication, and a shift in government policy towards an emphasis on food crop
production. It is unclear whether serendipity or particular individuals deserve credit for
recognizing this confluence. Regardless, irn 1982 the World Bank initiated Projet Centre-
Nord (PCN), a rural development project which «plicitly linked together these various
agencies.

Specific objectives for the PCN included improving the area’s rural
infrastructure, and strengthening of the agricultural research and «tension network,
particularly its management. The project, which targeted the cotton gowing regions of
the current Far North Province and the Mayo Louti Department of the North Province,
was implemented by SODECOTON under a World Bank contract.

Although the timing of PCN was fortuitous, the selection of SODECOTON as
the project’s managing entity probably contributed more to the project’s success. As a
business concerned with profit and the efficient use of its resources, SODECOTON’s
internal accountability and incentive structure provided the viable and relatively strict set
of checks and balances required for successful project implementation. Further,
SODECOTON’s functioning network of extension agents and input suppliers for cotton
farmers was relatively easy to «pand to food crop production.

The PCN era is most noteworthy for the successful forging of a multidisciplinary,
multifaceted effort for developing the region’s agicultural sector in general, and food

crop production in particular. Prior to PCN, each agency in the region functioned

12A
independently of each other. Although the shift in government policy” set the stage for

more collaboration, the PCN acted as the catalyst that molded food crop research and
development in northern Cameroon into an integated, interactive system.
5.1.4 Late 19803 - Restructuring

In 1987, the Cameroonian government declared a national economic crisis,
precipitated by the dramatic decline in world prices for all of Cameroon’s major «port
crops and by the burden of foreign debt. Although these problems were real, much of
the distress that characterized the government’s decrees about the crisis was simply a
smoke screen to cover the implementation of IMF-recommended austerity measures.
Fearing the political ramifications of restructuring, government chose a conveniently
construed scapegoat—the “economic crisis“.

The research-«tension system fared poorly in these tighter budgetary times.
Program budgets, salaries, and subsidies were all curtailed or eliminated. For IRA,
budget outlays for operating «penses dwindled to zero. By 1990, nearly all research
operating costs were being paid for by donor projects, while staff salaries for civil
servants were often paid months late and scheduled pay increases were igrored. When
PCN was phased out in 1987, food crop extension became costly for SODECOTON,
mainly because the government was unable to reimburse SODECOTON for these
activities. As a result, SODECOTON began charging IRA for the parastatal’s
«penditures for on-farm research trials. For all of the supporting agencies—NCSM

project, MIDEVIV, Office Céréalier, MINAGRI-the crisis translated into much smaller

 

3"This shift is documented in the Fifth-Five-Year Economic, Social and Cultural
Development Plan (p. 834-836, 851) in which IRA-Maroua’s research center is mandated
to «pand its activities in food crop research and to adopt a systems approach.

125
operating budgets, liquidation of assets (eg., most government-owned vehicles were sold),

and a general reduction in activities.

Two other noteworthy changes were motivated, at least in part, by the economic
crisis and subsequent redefinition of participating institutions”. First, starting in 1989,
SODECOTON drastically reduced its guaranteed price for cotton, essentially pegging it
to the world price. Historically, in years when the world cotton price fell below the set
price, SODECOTON subsidized the price paid to farmers rather than lower the
guarantee. This shift in pricing policy has led cash-crop farmers in northern Cameroon
to look for alternatives. In regions of lower rainfall, anecdotal evidence indicates that
cowpea gain production is one promising alternative farmers are adopting.

A second change in the region was motivated by both governmental restructuring
measures and the coming in vogue within USAID of “privatization”. In 1991, the USAID
sponsored N CSM project was sold to Pioneer Seed. While this analysis does not
consider how this change in property rights has affected or will affect northern
Cameroon’s agicultural sector, the privatization of the only seed company in the region
bears witness to the «tent of the government’s retraction from its former hay days of

interventionism“.

 

”By 1991, nearly all donor and government institutions had signed “performance
contracts” with the central government. These contracts defined the roles,

responsibilities, and performance criteria for both the central government and the signing
institution.

“This shift in property rights provides an «cellent opportunity to compare two types
of institutional arrangements. The NCSM project functioned for 15 years (1976 to 1991)
and data on its performance are available. Once Pioneer Seed has been established for
several years, this case seems to offer fertile gound for future inquiry.

126
5.2 Key Institutional Linkages, 1979-1987

Chapter 4 of this thesis estimated the net benefits of cowpea and sorghum
research and «tension. Using an IRR criterion, the chapter’s conclusions indicate that
the development of improved W and sorglnum technologies was relatively
”successful", particularly for the case of cowpea. Yet, these conclusions beg the question,
."Why were the progams successful?" The discussion that follows addresses this
fundamental question.

Analysis of key institutions, and their inter- and intra-relationships partially
«plain how "successes” were achieved in northern Cameroon. linkages within and
between such institutions as IRA, SODECOTON, and donor projects (eg., Bean/Cowpea
CRSP, SAFGRAD J.P. 31, NCRE and NCSM projects) proved critical to achieving
positive rates of return. The fact that an integated rural development project, PCN, was
implemented, in part, for the explicit purpose of linking together these institutions seems,
in hindsight, especially fortuitous.

Three insights are particularly clear from this analysis: (1) linkages within the
research-«tension system were critical; (2) linkages between the system and
international research institutions were equally important; and, (3) government
agicultural policies influence the system’s performance.

5.2.1 Linkages within the local system

Numerous efi'orts were made within the research-«tension system to link
together all of the “pieces” of the development ”puzzle". For example, the PCN made
investments to improve IRA’s management practices, hiring a coordinator to oversee the
agonomy research progam. His responsibilities included creating and maintaining links

between SODECOTON and IRA staff, which proved essential for the management of

127
off-station research (at research substations and in farmers’ fields). The coordinator’s

efi'orts facilitated information flows and fostered collaboration between IRA and
SODECOTON, and among each of IRA-Maroua’s commodity-based research units and
independent donor projects. Second, regularly scheduled stafl' meetings, organized by the
IRA-Maroua station director, provided an opportunity for interdisciplinary interaction
among researchers and staff. A third «ample was an annual planning meeting at which
each research unit presented the previous year’s results and the coming year’s research
agenda. Participants included representatives from SODECOTON, MINAGRI and
various NGO projects, as well as local farmers—all of whom were encouraged to provide
their input and evaluation of the planned research agenda.

These linkages among actors involved in the research-extension system enhanced
the technology development process in northern Cameroon in two key ways. First,
geatcr information flows served to inform system participants and proved an effective
means of identifying farmer constraints and setting the research agenda. For «ample,
as a consequence of this process, the mpea research agenda shifted from a primary
focus on producing high gain yields to addressing post-harvest storage constraints. This
shift was significant since post-harvest losses are now considered to be the largest
constraint to higher adoption of the already extended improved cowpea varieties.
Second, the linking of SODECOTON to the research system proved to be critical in the
overall performance of the system. SODECOTON, with its input distribution system and
500 to 1000 «tension workers, provided a conduit for both the «tension of technologies
and feedback from the farm to researchers. In turn, researchers knew that as they

developed appropriate technologies, a system was in place, ready to widely diffuse these

128

innovations. Knowing this proved to be an important motivating element for IRA’s
research staff.
5.2.2 Linkages beyond the local system

Linkages, via donor projects, between the agricultural research system and
international agicultural research centers (IARCs) also enhanced the technology
development process in northern Cameroon. Multilocational varietal screening trials
were organized at the international level by either IITA, SAFGRAD, ICRISAT, or the
Bean/Cowpea CRSP, and then implemented at the local level by either the CRSP,
SAFGRAD J.P. 31, or by the NCRE project. These trials beCame an important source
of alternative cultivars. Most of the varieties that were «tended to farmers as part of
the "improved” technology packages were actually introduced varieties first identified as
appropriate for the area through the international varietal screening trials. Hence,
IARCs and other international networks (CRSPs and regional projects), by collecting,
maintaining and distributing germplasm, acted as important catalysts for the agicultural
development process in northern Cameroon.

Further, donor projects in northern Cameroon had the capacity to access other
resources beyond those available to the national system, since all of the projects were
directly linked to international networks. This access clearly enhanced the performance
of the research-«tension system. Projects were able to provide, in addition to
introduced varieties, links to other research activities in the region, logistic support for
on-going research in Cameroon, and access to a network of other researchers who could

provide additional feedback relevant to the work being conducted by IRA-Maroua.

129
5.2.3 Government Policies

From 1979 to 1987, the Cameroonian government played a very limited role in
the agicultural sector of northern Cameroon. The ineffectiveness of MIDEVIV,
FONADER, Office Céréalier, and MINAGRI’s «tension system are all documented in
earlier sections of this thesis. Speculating on how the research-«tension system would
have performed under a difierent set of government policies is, at best, difiicult.
However, one issue merits comment. While farmers connected to SODECOTON’s
system of «tension and input delivery are much more likely to adopt improved
technologies, cotton farmers represent perhaps as few as 36% of all farmers in northern
Cameroon“. Hence, the adoption of technologies is dependent, in part, on which and
how many farmers are served by SODECOTON’s system. Had the extension and input
delivery system served a wider range of clientele, it is likely that the adoption of cowpea
and sorghum technologies in northern Cameroon would have been higher. However, it
is uncertain whether the benefits from attaining a higher adoption rate would
compensate the additional costs of establishing an «tension system which served a

broader constituency.

5.3 Distribution of Benefits
There is little documentation on the distribution of benefits from the
development and subsequent adoption of improved com and sorghum technologies in

northern Cameroon. Twa sources that gave some consideration to differentiated impact

 

“For the PCN region, the World Bank estimate was 36%, as reported in the 1980
PCN project paper. In 1991, during interviews conducted for this research, key
informants estimated that from 40 to 70% of the farmers in northern Cameroon
cultivate cotton.

130
among goups are: (1) data reported by Johnson on differences between cotton and
non-cotton farmers and (2) data reported by Wolfson on gender differences in cowpea
production and storage.

Johnson reports that “cotton sales dominate farm revenues in the Far North.
The mean annual revenue for a cotton-gowing family is 83,000 fcfa and for a non-
cotton-gowing family is 26,800 fcfa (p. 48, 1987)." He also notes that the only other
important sources of income for these farmers are ofi-farm and livestock revenues.
Given the research-«tension system’s dependency on SODECOTON, many of the
benefits of research were probably captured by cotton farmers, particularly early on in
the adoption cycle. This indicates that initially the beneficiaries, by income strata, were
probably the more aflluent farmers in the region. Given the improved cowpea
technology’s dependency on insecticide usage, this bias may still continue. With
sorghum, since the improved technology is an open-pollinated variety that did not
depend on a complementary technological package, lower income farmers probably also
captured some of the benefits of 835 as the technology was diffused.

In 1989, Wolfson, through work with the Bean / Cowpea CRSP, surveyed 112
households in the principal cOWpea gowing regions of northern Cameroon. Although it
is unclear as to the representativeness of Wolfson’s sample of cowpea farmers, her
results do indicate distinct gender differences in the production of W For «ample,
she notes:

Eiglnty-seven percent of the farmers produced cowpeas primarily for home

consumption. [Yet,] there was an association between the primary purpose of

cowpea cultivation in a household and the gender of the producer. When women
were responsible for production, the primary purpose was always for home
consumption (although some of the crop might get sold). In 17% of the

households in which men were involved in production either as sole producer or
cooproducer, the primary purpose was for sale...Women gew their cowpeas

131

intercropped with peanuts or sorglnum whereas men more frequently gew their
cowpeas in pure stands (p. 1-2; 1990).

One of Wolfson’s conclusions was that since women sell some of their cowpea crop,
changes in cowpea technologies could affect women’s access to this source of cash
income, indicating a need for researchers to be sensitive to this distributional change.

Based on Wolfson’s findings, the improved conea technologies probably favored
men, since the new system required monocropping and the use of insecticides. Wolfson
reported that both of these practices were found to be more prevalent with men. On the
other hand, cowpea production in general was reported to be more important to women,
implying that at least some of the benefits resulting from improvements in W
production are likely to be captured by them.

More conclusive discussions on distributions of benefits between income strata
and genders are limited, and other distributional issues (eg., differences between rural
producers and urban consumers, trade-offs between current and future generations) are
not «plored due to data constraints. Yet, because cowpea and sorghum are gown in
one of the poorest regions of Cameroon, the new technologies have enhanced the
welfare of these producers, vis-a-vis farmers in the higher rainfall, more well-endowed

regions of the country.

 

CHAPTER 6 SUMMARY AND CONCLUSIONS

This chapter is divided into four sections. The first summarizes the research
findings of the benefit-cost analysis, noting important issues affecting their interpretation.
The second compares the two rates of return estimated for the sorghum and cowpea
progams, reviewing possible causes for differences between the returns. The third
section summarizes the chapter (and thesis) with a succinct listing of key lessons learned
during this study. The final section looks beyond this work by considering the potential

implications for institutionalizing impact assessment.

6.1 Reported Rates of Return

During the period 1979 to 1986/87, investments in the cowpea and sorghum
research progams and «tension systems of northern Cameroon earned positive rates of
return. The estimated internal rate of return (IRR) for cowpea research and «tension
is 15%. The estimated IRR for sorghum research and extension is 1%. These two IRR
estimates do not stand on their own. Rather, their interpretation require a clear
understanding of both the assumptions and data supporting the IRR estimates, including
which costs and benefits were included/«eluded, which prices (real or nominal) were
used, and how qualitative issues such as institutional relationships and the distribution of
benefits were incorporated.
6.1.1 A critique of IRR Estimates

The rates of return are estimated for specific sets of investments and resulting

benefits. The delineation of which costs and benefits to include in the benefit-cost

132

133

streams was based on an iterative process that attempted to pair investments to their
corresponding benefits.

Investments in research that were included in the cost streams were limited to
only those commodity progams directly linked to the development of the improved
technologies. This delineation cuts both ways. By including all costs of a commodity-
based research progam, some costs (eg., the breeding progam irn the case of sorghum)
were included in the cost streams even though they did not contribute directly to the
improved technology's development. On the other hand, the «istence of a cadre of
researchers at the Maroua center provided support and interaction which enhanced the
overall performance of all of the individual progams. Would the new technologies have
been developed if there had «isted only a cowpea or a sorglnum progam? The question
is unanswerable, although limiting the cost streams to include only individual progam
costs and not aggegate system costs—as was done in this study—implies an answer in the
afiirmative.

A similar issue applies to «tension costs. The costs of the entire «tension
system were not included in the cost stream. Rather estimates were made of the
proportion of overall costs directly attributable to the specific commodity. It is uncertain
that an independent sorglnum (cowpea) «tension progam with a $200,000 (3 15,000) per
annum budget would have had the same impact as a $2.5 million per annum cotton and
food crop «tension progam. Yet, with this study, by partitioning out portions of the
food-crop component of SODECOTON’s total extension progam, these assumptions

were implicitly made.

 

134

Other costs which were «eluded from the analysis include investments made in
supporting industries, particularly the seed multiplication project and in supporting
research (eg., IITA’s costs for maintaining and «tending germplasm).

Furthermore, only benefits that increased the net market value of on-farm
production, which resulted from the adoption of the improved technologies, were
included. This «eluded a number of benefits, most of which were empirically difiicult to
quantify. For «ample, both of the improved W and sorglnum varieties are drouglnt
tolerant, contributing to improving the household food security of hundreds of thousands
of rural households in northern Cameroon. Yet, this benefit is not incorporated into the
benefit-cost analysis. Other «eluded benefits include the value of enhanced human
capital from project-trained host-national researchers, and the contribution of the
sorghum and mpea research efforts to the general pool of knowledge pertaining to
agicultural development in West-Central Africa. ‘

These assumptions and delineations about costs and benefits do not invalidate
the estimated IRRs. Such assumptions are required in order to carry out benefit-cost
analysis, but an awareness of which investments and which benefits netted internal rates
of return of 15 and 1% is necessary to put these numerical values in proper perspective.
6.1.2 Are these financial or economic returns?

Estimates of financial returns are based on nominal prices and expenditures
valued at purchased price, with no adjustments made for inflation, opportunity costs, or
over/undervaluation of local currency. When all of these adjustments are made, ROR
estimates represent economic returns. The «ception is when there is no inflation,

shadow «change rates equal market rates, and the opportunity costs of farm produce is

 

135
equal to its market values. In this case, estimated rates of return represent both the ‘

financial and economic returns.

Since no adjustments were made for inflation or the opportunity costs of the
resources included in the benefit-cost streams, the reported IRRs are the financial
returns to the investments in research and «tension. However, it is uncertain that
adjusting for inflation was necessary to represent the economic situation in northern
Cameroon. Given the «treme fluctuations in supply and prices in the region’s food-crop
markets, price trends indicating inflationary changes in the price level were not
discernable and may not exist.

Further, over the entire period of time framing the analyses, the «change rate
between the fefa and the French franc was fixed at 50 fcfa per one French franc. Thus,
fluctuations in the value of the fcfa simply reflected changes in the «change rate
between the US dollar and the French franc. Overvaluation of an «change rate that is
implicitly floating is only possible if certain conditions exist, such as differences in
purchasing power between trading partners, or unsustainable «ternal imbalance of
payments and trade distortions. No published data were found that determined if
and / or to what degee Cameroon’s currency was overvalued during the full period
considered, although key informants in Cameroon estimated that, in 1991, the currency
was overvalued by 40%. It is uncertain whether any overvaluation existed during the
1979 to 1986/87 period. Hence, no devaluation was deemed necessary during the base
runs, although assumptions about overvaluation were tested during sensitivity analysis,

resulting in lower IRR’s whenever the fcfa was devalued.

136

Finally, sorghum and cowpea are produced primarily for home consumption with
small surpluses marketed locally. Since these goods are not usually «ported, their
economic value (i.e., opportunity cost) is the market price they command.

If, in fact, there was no inflation in northern Cameroon, no overvaluation of the
Cameroonian currency, and the opportunity costs of farm inputs and outputs are valued
at their market price, then the estimated IRRs represent both the financial and
economic returns to research and «tension.

6.1.3 Qualitative Issues

Qualitative analyses seek to «plain why an investment had high, low or negative
returns. In the case of northern Cameroon, evidence indicated that certain institutional
relationships were critical to the attainment of positive returns since they acted as
catalysts for the agicultural development process. First, explicit linkages within the
research-«tension system, particularly between SODECOTON and IRA, were essential
to the system’s overall performance. These linkages established effective information
flows that helped identify farmers’ constraints and provided vital feedback from farmers
and «tension agents to researchers. Second, important institutional linkages were
established between the local research station and various international and regional
research centers and projects. Most of the improved varieties extended to farmers were
first tested in northern Cameroon as part of a network of international varietal screening
trials. In addition to introduced cultivars, these networks provided links to other
research activities in the region, logistic support for on-going research in Cameroon, and
access to other researchers who gave additional feedback relevant to the work being

conducted by IRA-Maroua.

137
Qualitative analyses also may help to «plain how returns are distributed. There

was some evidence that cotton farmers, particularly male cotton farmers, benefitted the
most from the development of the improved technologies. Two factors contributed to
this observation. First, IRA was closely linked to SODECOTON and depended on the
parastatal for support with on-farm research, for input delivery, and for the «tension of
improved technologies. Cotton farmers, a class of farmers dominated by males, were
more likely to be exposed to these innovations, have the means for adopting them, and
thus, most likely to benefit from their development. Second, in the case of W the
technologies «tended depended on such management practices as monocropping and

the use of insecticides, both of which were found to be more prevalent with men.

6.2 Differences in the Rates of Return
The significant difference in returns to the two commodity-based research ,

progams raises the obvious question, "Why?" Althougln a definitive answer to that
question is not possible, certain key characteristics differed between the two progams.

- First, the improved cowpea technology «tended to farmers represented a
completely new cropping system, while the improved sorghum technology was simply a
complement to traditional practices. Although cowpea is indigenous to northern
Cameroon, it was traditionally gown more as a garden crop, harvested for its leaves as
much as for its gain. The improved cowpea technology filled an existing need of
farmers in the region-an early maturing food crop to relieve hungy season food
shortages and to provide an alternative cash crop to cotton. On the other hand, under
normal rainfall conditions, S35 is just one more variety in the pool of over 1,800

accessions that have been identified in the region by the NCRE sorghum breeding unit.

 

138
S35 has enjoyed some success because it also addresses a need of farmers in the region-

a sorghum variety that is «tremely drouglnt tolerant. However, this need is not nearly as
predictable or regular as the needs met by the cowpea technologies. Hence, the most
obvious difference between the two progams was that cowpea research generated a
technology that netted benefits every year while the sorghum technology led to net
benefits only in drought years, whose frequency was estimated as one out of every three
years.

Second, given that this is a case study, little can be said about the general
appropriateness of funding screening progams versus breeding progams within research
projects. Yet, the higher returns were found with the W progam, which focused
entirely on varietal screening to "develop“ improved varieties. Even the success of the
sorghum progam depended not on a variety developed by its breeding progam but on a
variety identified in screening trials. Both cases imply higher returns were found for
screening activities. This conclusion is underscored by two important insiglnts. First,
screening progams are cheaper to a given national research progam because many of
the costs of generating an "improved" variety have already been incurred by other
projects and institutions. Second, the appropriateness of screening versus breeding
depends on its timing relative to the region’s overall state of development. Screening is
likely to be most successful early in the life of a research progam. As a first pass at
introducing improved technologies, high yielding varieties developed for a wide range of
gowing conditions (eg., TVX 3236) will likely have positive returns. However, after
these initial benefits are captured, and as researchers gain a geatcr understanding of the

constraints faced by farrrners within a specific region, breeding progams may be required

139
to address constraints which cannot be met with borrowed technologies (eg., cowpea

breeding progam established in 1988, targeting, in part, bruchid tolerance).

Third, another difference between the two commodities is that cowpea is a
relatively minor food crop, whereas sorglnum dominates the region’s cropping systems.
A priori, conventional wisdom suggests that the largest impact would result from
research targeted to the major crop. But is this so? In the late 1980s, the ma
market was poised for considerable «pansion. SODECOTON had cut its price subsidy
for cotton, leaving cash-crop farmers looking for alternatives. Corvpea had a competitive
advantage in production (drought tolerance) and in consumption (affordable protein
source), and could be readily sold in local markets, making it a viable alternative to
cotton. The change in cotton price represented an institutional shift iii the incentive
structure faced by farmers, which may «plain some of the relative "success“ of cowpea
technologies. Hence, the incentives that farmers face may influence the impact of
research more than the relative share a given commodity represents in a region’s
cropping system.

Fourth, the relative difficulty of the problems addressed by the two research
progams may also «plain some of the differences in the returns. Sorghum has I
presented a formidable problem to researchers throughout West and Central Africa for
over thirty years. Low returns to sorghum research, though undesirable, may simply

reflect long-run historical trends.

140

6.3 Summary of Key Lessons

The following is a succinct listing of the most salient conclusions of this study.

These conclusions present several insights on assessing the net benefits of technological

innovations in agiculture, particularly in a Sub-Saharan setting.

1.

When assessing impact, three issues need to be addressed: what were the
returns; why did the investments earn the returns that they did; and, how
were the returns distributed among the various segnents of society?

Institutions have a major impact on the rate of return since they influence

 

the performance of the research-«tension system, determine the
incentives faced by farmers, and define the means by which agicultural
transformation can be achieved. l
Since available data influence the assessment methodology employed,
issues of data availability and reliability are fundamental to benefit-cost
analysis.

Higher RORs are possible if costs can be reduced through cost sharing or
borrowing technologies. Examples from this study include insecticide
sprayers used by cowpea farmers (i.e., farmers already owned the sprayers
and used them on their cotton), and introduced cultivars borrowed
through international screening networks.

Research topics (eg., commodities, farmer constraints) differ in terms of
both potential for impact and in the degee of difficulty that they pose to
the researchers attempting to generate technological innovations. Hence,
selection of the constraint to research can influence the ROR as much as

the actual research that follows.

141
6.4 Institutionalizing Assessment Methodologies

Underlying the issues of agicultural technology assessment is a more
fundamental issue concerning the costs and benefits of data. If impact assessment is to
be institutionalized within Sub-Saharan NARSs, financial resources must be comnnitted
to generate appropriate data to support these analyses. This study confirmed that
administrators, plant breeders, and agonomists are not well versed in the methods and
scope of data collection necessary for economic analysis. Assessing the economic returns
of projects and/ or research-«tension systems is highly dependent on specific types of
data. Historically, these data have not been collected or have been given a low priority
in the research agenda.

The following list is provided as a field guide for project managers and
researchers as to the minimum data needed if impact assessment is to be conducted:
yields, prices, area in production, indicators of adoption, project «penditures, and
economic indicators.

Data are needed for yields associated with all major crop management practices
observed in farmers’ fields and for all of the products harvested (cg, gains, leaves for
food, stalks for fodder) under each set of practices. This includes yields under
traditional practices, and yields under various levels of adoption (eg., variety adopted
verses variety and insecticide adopted). In general, plant breeders and agonomists are
hesitant to estimate an “average yield“ for a given set of crop management practices,
recognizing the plethora of factors that influence yield. To obtain these data, research
progams must establish protocols for representatively sampling farmers’ yields under the

major systems observed in the target research environment.

142
Input and output prices are absolutely necessary, if the dollar values of the

benefits and impacts of an improved technology are to be estimated. Further, input
costs must be described (e.g., formulation, brand) for each crop management practice,
since it is impossible to estimate farmers’ input costs if the list of specific inputs is
unknown. With respect to output prices, the need to capture price fluctuations and long
run price trends requires periodic collection of price data (eg., monthly or, when
possible, weekly prices over several years). Prices also vary across space. This implies
that data must be collected on prices received by farmers, prices paid by urban
consumers, and border prices of tradable goods. Finally, since plant by-products (eg.,
leaves for food, stalks for fodder) have economic value and may be marketable, prices
must be collected for these markets as well.

Census data such as area harvested by commodity, area encompassed by the
research-extension system, and number of farmers in these areas are fundamental to
assessing impact. Area in production data at the farm level (eg., average farm size, land
allocations between commodities) are also important, especially if “people-level” impact
assessment is an objective.

Indicators of adoption are necessary since these data are required to estimate
benefits from the development of improved technologies. Seed sales, farmer surveys and
market surveys are all possible proxies for adoption. If farmer surveys are used, then
enquiries must go beyond the bimodal observation of whether or not the farmer is using
the improved technologies. Questions must ask about the production mix (total adoption
versus adoption on only a portion of the land cultivated to the given crop) since this

information is needed to estimate total area under the improved cropping practices.

143
Finally, farmer surveys must follow sampling procedures that guarantee that these data

are collected from a representative sample of the producers in the targeted area.

Records are almost universally kept of project and station «penditures. Simple
annual summaries of these records would geatly facilitate impact assessment efforts. An
annual one-page summary of «penditures on staff salaries, operating costs,
administration costs, training costs, and capital improvements would be sufficient for ;
benefit-cost analysis. If each entity within a research-«tension system (eg., IRA,
SODECOTON Extension Services, NCRE) prepared these summaries, then the
estimation of cost streams for benefit-cost analysis would be limited to compiling these
records.

The last type of data needed for economic analysis-indicators of inflation, I
«change, interest, and labor-wage rates—may be beyond the scope of most
agonomic/production-oriented research projects. Information on inflation and exchange
rates may be available in-country from the IMF or World Bank. If so, project managers
can glean this information at low cost. Local interest rates and labor wage rates, like
most socio-economic data, can be collected by surveys and/ or monitoring of local
markets.

Finally, institutionalizing impact assessment depends not only on data collection,
but also on trained social scientists to analyze these data. Additional investments within
NARS would be necessary to develop human capital capable of conducting economic

analyses and impact assessments.

APPENDIX

145

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table A2 Cowpea, Hectares Harvested, Far North Province, Cameroon for
Years 1972/73 to 1989/90.
Area Harvested (Ha) Cowpea’s Share
Year Cowpea All crops (%)
1972/73 67,006 438,000“ 1.53
1973/74 4,900 429,000 1.14
1974/75 4,000 382,000 1.05
1975 /76 138,000 775,000 17.81
1976/77 149,000 689,000 21.63
1977/78 86,000 636,000 13.52
1978/79 156,000 649,000 24.04
1979/80 159,000 696,000 22.84
1980/81 28,000 543,000 5.16
1981/82 40,069 445,514 8.99
1982/ 83 22,329 443,689 5.03
1983/84 24,345 337,925 7.20
1984/85 66,254 463,257 14.30
1985/86 42,252 571,313 7.40
1986/87 45,174 ' 633,483 7.13
1987/88 45,360 617,055 7.35
1988/89 36,785 598,659 6.14
1989/90 32,120 618,436 5.19
Source: MINAGRI, various national and provincial reports.

 

“Data available prior to 1981/ 82 are for the former North Province which is now
divided into the 3 northern provinces of Adamaoua, North and Far North. To convert
these data to Far North province ”equivalents”, total cropped area was multiplied by a
factor of 0.64 and a factor of 0.83 was used to convert total cowpea area. These factors
were based on the relative shares reported in the Agicultural Census figures, years
1984-89.

BIBLIOGRAPHY

 

Ii

BIBLIOGRAPHY

Akino, M. and Y. Hayami. 1975. “Efficiency and Equity in Public Research: Rice
Breeding in Japan’s Economic Development.” American Journal of Agicultural
Econorrnies, 57:1-10.

Anderson, Jock R. and Robert W. Herdt. 1990. ”Reflections on Impact Assessment”in
etosor ao R cSe st ts th
t al esea . Vol. 2: Assessing Impact of Agicultural Research. aR.
G. Eeheverrfa, ed. ISNAR, The Hague.

Busch, Lawrence, Paul Anamosa, and Tim Schilling. 1989. Cameroon Country Report:
Agricultural Research Impact Indicators. Management Systems International,
Washington DC

Capalbo, Susan M. and John M. Antle. 1989. ”Incorporating Social Costs in the Returns
to Agicultural Research. " American Journal of Agicultural Economies,
71: 2. 458-463.

Conniff, KL. 1987. ”Cowpea/Sorghum Intercrop Systems in Semi-Arid Botswana:
Assessment of Yield, Growth, and Water Use.“ PhD dissertation, Colorado State
University, Fort Collins, Colorado.

Contant, Rudolf B. and A. Bottonnley. 1988. Priority Setting in Agricultural Research.
ISNAR, The Hague, Netherlands.

Economic Intelligence Unit. 1989. World Outlook 1989. Business International Limited,
London.

Eeonorrnic Intelligence Unit. 1990. Cameroon, CAR, Chad Country Report. Business
International Limited, London. N0. 2.

Economic Intelligence Unit. 1991. Cameroon, CAR, Chad Country Report. Business
International Limited, London. N0. 1.

Eeheverria, R. G. 1990. "Assessing the Impact of Agicultural Research" nn Mfg;
os ea c ste t a
Agricultural Research. Vol. 2: Assessing Impact of Agicultural Research. R. G.
Echeverria, ed. ISNAR, The Hague.

146

 

z-I'A‘ r5. . Q". ' cc 10 a
.2

147

Eicher, Carl K. and John M. Staatz. 1984. ”Agicultural Development Ideas in

Historical PerSPective' in W Ed by
Eicher and Staatz. John Hopkins University Press, Baltimore, MD.

Evenson, RE, RE. Waggoner, and V.W. Ruttan. 1979. “Economic Benefits from
Research: An «ample from Agiculture." Science. 205:1101-1107.

Evenson, RE. 1989. 'Spillover Benefits of Agicultural Research: Evidence from US.
Experience.“ American Journal of Agicultural Eeononnies, 71:447-452.

Evenson, RE. 1991. Notes on the Measurement of the Economic Consequences of
Agricultural Research Investmena. Notes prepared for a conference on
international agicultural research, Cornell University, May 17-19, 1991.

Gittinser. JP 1982. WWW Economic
Development Institute of the World Bank. John Hopkins University Press,
Baltimore.

Grilihes, Z. 1958. Research Costs and Social Returns: Hybrid Corn and Related
innovations. Journal of Political Economy, 66:419-431.

Griliches, Z. 1964. ”Research Expenditures, Education, and the Aggegate Agicultural
Production Function.” American Economic Review, 54: 961-974.

Haque, Enamul, A.K., G. Fox and 6.1.. Brinkman. 1989. ”Product Market Distortions
and the Returns to Federal Laying Hen Research in Canada.” Canadian Journal
of Agicultural Economics, 37 :29-46.

Henry de Frahan, Bruno. 1990. ”The Effects of Interactions Between Technology,
Institutions, and Policy on the Potential Returns to Farming Systems Research in
Semi-Arid Northeastern Mali.” PhD Dissertation, Department of Agicultural
Economies, Michigan State University.

Holtzman, J .S. 1986. "Rapid Reconnaissance Guidelines for Agicultural Marketing and
Food System Research in Developing Countries." Michigan State University
International Development Papers, Working Paper No. 30. Department of
Agicultural Economics, Michigan State University, East Lansing Michigan.

Horton, D. 1990. ”Assessing the Impact of International Research: Concepts and
Challenges. in et 0 S te

WW Vol 2 Assessing Impact of
Agicultural Research. R. G. Echeverria, ed. ISNAR, The Hague.

Imung', Jasper K. and Norman N. Potter. 1983. "Nutrient Contents of Raw and Cooked
COWpea Leaves.” Journal of Food Science, 48:1252-1254.

Jaeger, William 1c 1987. US. Aid to Cameroon: 1n- Impact on Agriculture and Rural
Development. MADIA Study, The World Bank, Washington DC.

 

148

Jahnke, Hans E., Dieter Kirschke and Johannes Lagemann. 1987. The Impact of
Agricultural Research in TlopicalAfi-ioa'A Study of the Collaboration between the
International and National Research Systems. CGIAR Study Paper No. 21, The
World Bank, Washington, DC.

Karnuanga, M. 1991a. Les Coats de Produca'on des Culture Vivrieres dans l’Extreme-Nord
TLU-Maroua Working Paper No. 1, IRA-Maroua, Cameroon.

Kamuanga, M., J.T. Russell, C. Njomaha, and M. Fobasso. 1991b. Les Systemes de
Productiondansl’Exaeme-Nordetlesconsaaintesalfiméliomtiondela
Productivité. Résultats d’une Enquete Exploratoire. TLU-Maroua Working Paper
No. 2, IRA-Maroua, Cameroon.

. n.5,: n _ .eu...

Kamuanga, M. and Martin Fobasso. 1991c. Role of Farmers in the Evaluation of an ,
Improved Variety: The Case of S35 Sorghum in Northern Cameroon. Paper l-
Presented at the 1991 Farming Systems Symposium, East Lansing, Michigan.

 

Karanja, Daniel D. 1990. 'Ihe Rate of Return to Maize Research in Kenya: 1955-88."
MS Plan B paper, Department of Agicultural Economies, Michigan State
University.

Kitcln, Laurie. 1990. Preliminary Results of a Survey of Cowpea Farmers. IRA-Maroua
Legume Progam, Cameroon.

little, I.M.D. and LA. Mirrlees. 1991. ProjectAppraisal and Planning TWenty Years On,
Proceeding of the World Bank Annual Conference on Development Econonnies
1990. The World Bank.

Lu, Yao—chi, L. Quance, and Chun-lan Liu. 1978. “Projecting Agricultural Productivity
and Its Economic Impact.“ American Journal of Agicultural Economies, 60:976-
984.

Martin, F. 1988. "Budgets de Culture au Sénégal.’ Michigan State University
Development Papers, Reprint No. 28F. Department of Agicultural Econorrnies,
Michigan State University, East Lansing, Michigan.

Mbuagbaw, T.E., R. Brain, and R. Palmer. 1987. Wm Longnan
Group UK Limited, Essex, England.

Mellor. 1W. 1966. Women; Cornell University
Press, Ithaca, NY.

Mehta, P. N. 1971. ”The Effect of Defoliation on Seed Yield of Cam and Analysis
of the Leaf Harvest for Dry Matter and Nitrogen Content. " Acta Horticulturae
21:167-172.

149

Ministry of Plan. Undated. Demo 87: 7.5 Million in 1976, We are 10.5 million
Inhabitants in 1987. SOPECAM, BR 1218, Yaounde, Cameroon.

Nagy, J .G. and W.H. Furtan. 1978. Economic Costs and Returns from Crop
Development Research: The Case of Rapeseed Breeding in Canada." Canadian
Journal of Agricultural Economics, 26:1-14.

National Directorate of the Agricultural Census/MINAGR11987.1W
. z - _ , _ - Volume 2A.
MINAGRI, Yaounde. Ef‘

 

National Directorate of the Agicultural Census/MINAGR11987.
~ ' . . , - .Volume 2B.

 

MINAGRI, Yaounde.

- R-H‘L" '

Njomaha, C. and M. Kamuanga. 1991. Le Sorgho de Saison Séche en Milieu Paysan de
I’Extreme-Nord. Productivité et Constraintes. TLU-Maroua Working Paper No. 3,
IRA-Maroua, Cameroon.

 

Oehmke, James F., L. Daniels, J. Howard, M. Maredia, and R. Bemsten. 1992. The
Impact ongricultural Research: A Review ofthe Ex-PostAssessment Literature
with Implications for Afiica Department of Agicultural Econonnies Stafl Paper
No. 92-38, Michigan State University, East Lansing, MI.

Perez, AT. 1979. Plant Exploration in Cameroon, October to December 1979. Mission
Report, IITA.

Pradey and Rosenboon.1989.

 

Press, Cambridge, England.

Pradey, Phillip G. and Stanley Wood. 1991. Targeting Research by
Environments. Paper prepared for the World Bank conference on ”Agicultural
Technology: Current Policy Issues for the International Community and the
World Bank“, Airlie House, Virginia.

Russell, LT. 1991. "Yield Stability of Pure and Mixed Stands of Sorghum Varieties in
Northern Cameroon.” A PhD dissertation, University of Florida, Gainsville,
Florida.

Salinger, B. Lynn, and J. D. Stryker. 1991. Exchange Rate Policy and Implications for
Agricultural Market Integration in West Africa. Associates for International
Resources and Development, Cambridge, Massachusetts.

Schmid. AA 1987 W
W Praeger Publishers, New York, New York.

150

Schultz, T.W- 1953- Walkers. McGraw-Hill. New
York.

Seré C. and LS. Jarvis. 1990. 'Ihe Betting Line on Beef: Ex ante Estimations of the
Benefits of Research on Improved Pastures for the Latin American Tropics“ in
U‘I‘° ‘ .H1'1I9‘I'x'NAA; ISb A°I 1.1-‘uLIQé.-AX.I'.I' ”'2‘!
W. Vol. 2: Assessing Impact of Agicultural Research. R.
G. Echeverria, ed. ISNAR, The Hague.

Singh, S. R. and K. O. Rachie. - '
Wiley and Sons, New York, 1985.

 

Ta’Ama, Mofi. 1984. Provisional Report on Cowpea Baseline Data Survey in Northern
Cameroon. IRA-Maroua, Cameroon.

Talleyrand, Henri, Titus N goumou Nga, and Anatole Ebete Mbeng. 1986.128:

I , , .
\flNAA'. I DA I . 'J I! nJ.I'II '1' ' At :.II_ :-I'. I

W IRA/NCRE. Garoua.

Thirtle, C. and P. Bottorrnley. 1988. 'Is Publicly Funded Agicultural Research
Excessive?“ Journal of Agricultural Economics, 39:99-111.

TLU. 1990. Synthése des Résultats de Recherche 1990, Section TLUflVCRE/IRA. IRA-
Maroua, Cameroon.

Wolfson, Jane L. 1990. Analysis of Cowpea Production and Storage Methodologies Used
By Small Farmers in Northern Cameroon. Bean/Corvpea CRSP Project Report,
Department of Entomology, Purdue University, West Lafayette, IN.

General References

Bean/Cowpea CRSP - Michigan State University. 1983. 1983 Annual Report, Technical
Summary. East Lansing, Michigan.

Bean / Cowpea CRSP - Mielnigan State University. 1983. 1983 Annual Report, External
Revrew Panel CRSP, East Lansing, Michigan. n

Bean / Cowpea CRSP - University of Georgia. 1983. Pest Management Strategies for
Optimizing Cowpea Yield in Cameroon: Annual Report. Athens, Georgia.

Bean/Cowpea CRSP - Michigan State University. 1984. 1984 Annual Report Part 1,
Technical Summary. East Lansing, Michigan.

 

Bean/COWpea CRSP - Michigan State University. 1984. 1984 Annual Report Part 2, 1.
External Review Panel. CRSP, East Lansing, Michigan. i

Bean/Cowpea CRSP - Michigan State University. 1984. Collaborative Research in the
International Agricultural Research and Development Network: A Case Study,
Progress Report of the Bean/Cowpea CRSP. East Lansing, Michigan.

Bean / COWpea CRSP - Michigan State University. 1985. Research Highlights. “A New
Look at the Importance of Cultivars in Cowpea Research: Evidence from
Northern Cameroon” by Dr. Mofi Ta’Ama. Volume 2 No. 5. East Lansing,
Michigan.

Bean/Cowpea CRSP - Michigan State University. 1988. 1988 Annual Report, Technical
Summary. East Lansing, Michigan.

Bean/Cowpea CRSP - Purdue University. 1989. 1989 Growing Season Results,
CRSP/IRA, Maroua Cameroon. West Lafeyette, Indiana.

Bean /Cowpea CRSP - Michigan State University. 1990. Report of the External
Evaluation Panel of the Bean/Cowpea CRSP for FY 90. East Lansing, Michigan.

Bean / Cowpea CRSP - Purdue University. 1990. Supplement to the 1990 Detailed Project
Annual Report, October 1, 1990--December 31, 1990. West Lafeyette, Indiana.

Boutrais. 1.. et al 1984. Wu. Collection
Mémmoires n°102, OSTROM, Paris.

151

152

Boyce, James K. and Robert E. Evenson. 1975. National and IntemationalAgricultural
Research and Extension Programs. Agicultural Development Council, Inc. New
York, New York.

Dangi, O.P., Richard Kenga, and Andre D. 1986. “Sorghum and Millet Improvement
Progam NCRE Project (IITA - USAID)", Annual Report - 1985. IRA/NCRE,
Maroua.

Delegation Provinciale de l’Agiculture de l’Extréme-Nord. 1986. Rapport Annuel
d'AcliVités: Exercice 1985-1986. DPAEN, Maroua. T

Delegation Provinciale de l’Agiculture de l’Extreme-Nord. 1987. Rapport Annuel
d'Activites: Exercice 1986-1987. DPAEN, Maroua.

Delegation Provinciale de l’Agiculture de l’Extreme-Nord. 1990. Rapport Annuel
ditlctivites: Exercice 1989-1990. DPAEN, Maroua.

 

Division des Enquétes Agoéconomiques et de la Planification Agicole (DEAPA).
1991. 1984 - 1989 NationalAgr-icultural Surveys: National and Provincial Results
(Provisional). Unpublished document.

Division des Etudes/MINAGRI. 1988. An Executive Guide to the 1984 Agricultural
Census, Part 1: Farm and Farmer Characteristics, Farming Methods and Practices.
MINAGRI, Yaounde.

Dromard, Philippe. 1986. Les Resultats de la Campagne Agricole 1985-86. Evaluation de
l’Impact de la SODECOTON sur les Paysans. SODECOTON, Garoua.

Dromard, Philippe. 1987. Rapport Final, Bilan de 3 Campagnes Agricoles dans le Centre-
Nord. Louis Berger International, Garoua.

Fanso, V..G 1989. : ' - - -
Colonial and Post-Colonial Periods. " MacMillan Publishers, Ltd., Hong Kong.

 

Fox, Glenn. 1987. ”Models of Resource Allocation in Public Agicultural Research: A
Survey." Journal of Agicultural Economics, 38:3:449-462.

Gouthiere, J. 1983. Centre de Recherches Agronomiques de Maroua: Antennes,
Experimentation Hors-antenne, Systémes de Culture 1982-1983. IRA, Maroua.

Gouthiere, J. 1984. Centre de Recherches Agronomiques de Maroua, Campagne 1983/84,
Projet Centre-Nord. IRA, Maroua.

Gouthiere, J. 1985. Centre de Recherches Agronomiques dc Maroua, Campagne 1984/85,
Projet Centre-Nora. IRA, Maroua.

Gouthiere, J. 1986. Rapport de Mise en Place de la Campagne 1985/86. IRA, Maroua.

153

Goutlniere, J. and G. N’gono. 1986. Campagne 1985/86 - Rapport Annuel. Projet Centre
Nora. IRA, Maroua.

Gouthiere, J. and G. N’gono. 1986. Campagne 1985/86 - RapportAnnuel. Projet Centre
Nord - Anrieces. IRA, Maroua.

Goutlniere, J. and G. N’gono. 1987. Centre de RecherchesAgronomiques de Maroua,
Campagne 1986/87, Projet Centre-Nord. IRA, Maroua.

Gouthiere, J. and G. N’gono. 1987. Campagne 1986/87. Projet Centre Nord - Annexes.
IRA, Maroua.

Gouthiere, J. 1987. Projet Centre-Nora: Bilan de 5 Arts d’Activités (1982/83 d 1986/87).
IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1980. Sorghum and Millet Regional Dial in
Northern Cameroon 1980 Season. SAFGRAD J.P. 31, IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1981. Rapport Préliminaire des Résultats de l’Essai
Varietal Avancé de Sorgho Realise par le Projet Pilate Ago-Pastoral de Mindif-
Moulvoudaye - Campagne 1981. SAFGRAD J.P. 31, IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1982. Rapport Préliminaire sur les Principawt
RésultatsAcquis etsurles Nouveaux Themes Prepares, a l’Occassion de la Reunion
d’Information et de Programmation entre l'Institute de la Recherche Agronomique et
la SODECOTON. SAFGRAD J.P. 31, IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1983. Rapport Analytique de Deux Essais Varietaux
Avancés de Niébé Realises par le ProjetAGRILAGDO a Karewa. SAFGRAD J.P.
31, IRA, Maroua.

Gwathmey, 'C. O. and M. T. Fobasso. 1983. Rapport Preliminaire sur les Résultats des
EssaisduProjetSAFGRADsurles CéréalesPreparepourlaDeuxiemeRéunionde
Programmation du Projet NCRE quise Tiendre a Douala en Janvier 1983.
SAFGRAD J.P. 31, IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1983. Projet SAF GRAD au Cameroun:
Programme didctivités Propose pour 1983/84. SAF GRAD J.P.31, IRA, Maroua.

Gwathmey, Owen and Martin Fobasso. 1983. Rapport Intén'maire d Mctivité, Campagne
1983. SAFGRAD J .P. 31, IRA-Maroua, Cameroon.

Gwathmey, Owen and Martin Fobasso. 1983. Rapport Preliminaire sur les Principaux
Résultats Acquis et sur les Nouveaux Themes Prepares. SAFGRAD J.P. 31, IRA-
Maroua, Cameroon.

 

154

Gwathmey, Owen and Martin Fobasso. 1984. RapportAnalytique des Résultats 1983 sur
le Niébé. SAFGRAD J .P. 31, IRA-Maroua, Cameroon.

Gwathmey, C. O. and M. T. Fobasso. 1984. Programme de Recherches Propose pour la
Campagne 1984. SAFGRAD J.P. 31, IRA, Maroua.

Gwathmey, C. O. and M. T. Fobasso. 1984. Bilan de Réalisation du Projet SAFGRAD au
Cameroun 1979-1983. SAFGRAD J.P.31, IRA, Maroua.

Hardaker, J. Brian, Jock R. Anderson and John L. Dillon. 1984. "Perspectives on P-
Assessing the Impacts of Improved Agicultural Technologies in Developing
Countries." Australian Journal of Agicultural Econonnies, 28:2 & 3:87-108.

IRA. 1979. Rapport Annuel 1979. DGRST-IRA, Yaounde.

IRA. 1982. Programme de Recherches 1981 - 1982. DGRST-IRA, Yaounde.

 

nt-

IRA. 1983. Resume 1982. IRA, Maroua.

IRA. 1984. Pluviométrie de la Zone Cotonniere du Nord Cameroun (20 postes). IRA,
Maroua in collaboration with SODECOTON, Garoua.

IRA-Progamme Légumineuses. 1985. Programme Légumineuses: Compte Rendu de la
Reunion du 07 au 08 Juin 1985. IRA, Maroua.

IRA. 1985. Syndiesis of 1985 Research Findings, Farming System Programme. IRA,
Yaounde.

IRA. 1985. synthesis of 1985 Research Findings, Cereals Programme. IRA, Yaounde.
IRA. 1987. I986 On-Farm Tests. TLU, IRA-Maroua, Cameroon.

IRA - Progamme Legumineuses. 1987. Le Point de la Recherche sur le Niébé au Centre
de Recherche Agronomique de Maroua, Cameroun. IRA, Maroua.

IRA - Progamme Légunineuses. Undated (1987). Cowpea Storage Research 1986-1987.
IRA, Maroua.

IRA and the Government of Cameroon. 1990. Performance Contract between the
Government of Cameroon and the Institute of Agricultural Research (IRA).
IRA/ RC, Yaounde.

IRAF. 1977. Principaux Résultats du Centre des Cultures Textiles et Vtvrieres de Maroua.
IRAF, Buea.

IRAT. 1967. L’IRAT au Service du Cameroun, Rapport de synthése de l’Année 1967.
IRAT, Yaounde.

155
IRAT. 1968. L’IRATau Service du Cameroun de année 1968. IRAT, Yaounde.

IRAT. 1970. Rapport D’Activite en Republique Fédérale du Cameroun. IRAT, Yaounde.
IRAT. 1974. Le Sorgho IRAT 55. IRAT, Maroua.

ISNAR and Pan-African Institute for Development—Central Africa. 1984. Improvement
of Agricultural Research Management in Cameroon. Report to the Delegation
General for Scientific and Technical Research of the United Republic of
Cameroon. R 19. The Hague, Netherlands. :l"

I

ISNAR. 1988. An Analysis of Structure and Management of die Institute of Agricultural 5.
Research (IRA) and the Institute of Animal Research (IRZ) of Cameroon. Report I
to the Ministry of Higher Education and Scientific Research of the Republic of E
Cameroon. R 28c. The Hague, Netherlands :

Jaeger, William K. 1987. US. Aid to Cameroon: Its Impact on Agricultural and Rural -
Development Report for the World Bank MADIA Study, Washington, DC. "

Johnson, Eric. 1987. Demand Estimation and Farmer Surveys for die North Cameroon
Seed Multiplication Project NCSM, Garoua.

Johnson, Jerry and Jean Nzoning. 1985. 1984 Sorghum Report of On-Farm Tests.
SAFGRAD J .P. 31 in Cameroon, IRA-Maroua, Cameroon.

Johnson, Jerry J. and M. T. Fobasso. 1986. Synthesis of Results for the 1985 Crop Season.
IRA/SAFGRAD, Maroua.

Johnson, Jerry and Martin Fobasso. 1988. 1987 On-Farm Testing. TLU, IRA-Maroua,
Cameroon.

Johnson, Jerry J. 1988. Final Report of On-farm Testing within IRA Maroua 1984-1987.
IRA, Maroua.

Johnson, Jerry L, J. R. Alldredge, S. E. Ullrich, and O. Dangi. 1992. "Replacement of
Replications with Additional Locations for Grain Sorghum Cultivar Evaluation."
Crop Science, 32:1:43-46.

Kamuanga, M., J. Russell, and M. Fobasso. 1990. Production systems and Farmers'
Constraints in the Mayo-Sava and Mayo-Tsanaga Departments of Extreme-North
Cameroon: Implications for Research. TLU Working Paper, IRA-Maroua,
Cameroon.

Lynch, Sarah G. 1991. ”Income Distribution, Poverty and Consumer Preferences in
Cameroon." Cornell Food and Nutrition Policy Progam, Washington, DC.

156

Minot, Nicholas. 1991. Impact of the Fertilizer Sub-Sector Reform Program on Farmers:
The Results of Three Farm—Level Surveys. Report prepared for the Agicultural
Marketing Improvement Strategies Project, Abt Associates, Washington, DC and
the University of Idaho /Postharvest Institute.

NCRE. 1983. Annual Report 1982. NCRE, Yaounde.

NCRE. 1983. Work Plan 1983. NCRE, Yaounde. '

NCRE. 1984. Annual Report 1983. NCRE, Yaounde.

NCRE. 1985. Annual Report 1984. NCRE, Yaounde.

NCRE. 1987. Annual Report 1986. NCRE, Yaounde.

NCRE. 1988. Annual Report 1987. NCRE, Yaounde.

NCRE. 1988. Annual Work Plan (March 1987 - Feb 1988). NCRE, Yaounde.

NCRE. 1989. Annual Work Plan (March 1988 - Feb 1989). NCRE, Yaounde.

NCRE. 1989. Annual Report 1988. NCRE, Yaounde.

NCRE. 1990. Annual Work Plan (March 1990 - Dec 1990). NCRE, Yaounde.

NCRE. 1990. Annual Work Plan (March 1989 - Feb 1990). NCRE, Yaounde.

NCRE. 1990. Annual Report 1989. NCRE, Yaounde.

NCRE. 1991. Annual Report 1990. NCRE, Yaounde.

NCRE. Undated. Sorghum and Pearl Millet Research Program in North-Cameroon.
NCRE, Yaounde.

NCSM. 1977. Rapport Synthétique des Activites du Projet Semencier Campagne 1976/77.
NCSM, Garoua.

NCSM. 1978. Rapport Synthétique des Activites du Projet Semencier Campagne 1977/78.
NCSM, Garoua.

NCSM. 1979. Rapport Annuel Campagne 1978/1979. NCSM, Garoua.

NCSM. 1980. Rapport d’Activités du Projet Semencier Campagne 1979/80. NCSM,
Garoua.

NCSM. 1981. Rapport Synthétique des Activites du Projet Semencier US-AID/RUC
Campagne 1980/81. NCSM, Garoua.

157

NCSM. 1981. Rapport ExhaustifdesActivites du Projet Semencier au Cours de la Premiere
Phase, Campagnes 1976/77 d 1980/81. NCSM, Maroua.

NCSM. 1982. Rapport synthetique des Activites du Projet Semencier US-AID/RUC
Campagne 1981/82. NCSM, Garoua.

NCSM. 1983. Rapport Syndietique des Activites du Projet Semencier US-AID/RUC
Campagne 1982/83. NCSM, Garoua.

NCSM. 1984. Rapport didctivités du Projet Semencier US-AID/RC Exercice 1983/84.
NCSM, Garoua.

NCSM. 1985. Rapport Synthétique des Activites du Projet Semencier US-AID/RC
Campagne Agricole 1984/85. NCSM, Garoua.

NCSM. 1986. Activities Report NCSM, January - June 1986. NCSM, Garoua.

NCSM. 1986. Rapport Syndiétique du Projet Semencier US-AID/RC Campagne Agricole
1985/86. NCSM, Garoua.

NCSM. 1987. Rapport Synthetique du Projet Semencier US-AID/RC Campagne Agricole
1986/87. NCSM, Garoua.

NCSM. 1988. Note de Synthese des Activites du Projet Semencier Vivrier - Nord de 1976 a
1988. NCSM, Garoua.

NCSM. 1989. Programme d’Activites du Projet Semencier Nord, Campagne 1988/89.
NCSM, Garoua.

NCSM. Undated. Getting Acquainted with the North Cameroon Seed Multiplication
Project NCSM, Garoua.

Nleuma. Martin, et al 1989 W
W ed. by Njeuma. Macmillan Publishers Ltd., Hong Kong

Progamme National de Vulgarisation et de Formation Agicoles. 1991. Rapport
Semestriel d ’Activité: .Iuillet - Decembre 1990. PNVFAEN/MINAGRI, Maroua.

Progamme National de Vulgarisation et de Formation Agicoles. 1991. Rapport Annuel
d’Activité 1990/1991; Programme didctions 1991/1992. PNVFAEN/MINAGRI,
Maroua.

Russell, John, and Martin Fobasso. 1989. 1988Annual Report, Testing and Liaison Unit,
Agronomy Research Center: TLU, IRA-Maroua, Cameroon.

SAFGRAD J.P. 31/Cameroon. 1980. Essai Varietal de Niébé (SARCVT) dans la Zone
Semi-Arid, Campagne 1980. Unpublished trial protocol, SAF GRAD, Maroua.

158

SAFGRAD J .P. 31 /Cameroon. 1980. Essai Entomologique des Niebés, Campagne 1980.
Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J .P. 31 /Cameroon. 1980. Essai Varietal des Sorgho a cycle Intermédiare,
Campagne 1980. Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J .P. 31/Cameroon. 1980. Essai Varietal des Sorgho a cycle Court, Campagne
1980. Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J .P. 31 /Cameroon. 1980. Essai Varietal des Mai's a cycle Intermédiare,
Campagne 1980. Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J.P. 31/Cameroon. 1980. Essai Varietal des Mats a Cycle Court, Campagne
1980. Unpublished trial protocol, SAF GRAD, Maroua.

SAF GRAD J .P. 31 /Cameroon. 1980. Essai de Commotion de Grains (Sorgho, Mats,
Niébé et Arachide). Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J.P. 31/Cameroon. 1981. Essai Varietal de Sorgho a Cycle Intermédiare.
Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J.P. 31/Cameroon. 1981. Essai Varietal de Sorgho a cycle Court.
Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J .P. 31 /Cameroon. 1981. Essai Varietal Regional de Sorgho d Cycle Long.
Unpublished trial protocol, SAF GRAD, Maroua.

SAFGRAD J .P. 31 /Cameroon. 1981. Essai Varietal Avancé de Sorgho Précoce.
Unpublished trial protocol, SAF GRAD, Maroua.

SAF GRAD J.P. 31 /Cameroon. 1981. Essai de Sorgho, 1981. Unpublished document,
SAF GRAD, Maroua.

Service de la Statistique Agicole/MINAGRI. Undated. Annuaire de Statistique
Agricoles 1976-1977. MINAGRI, Yaounde.

Service de la Statistique Agicole/MINAGRI. Undated. Annuaire de Statistique
Agricoles 1978-1979. MINAGRI, Yaounde.

Service de la Statistique Agicole/MINAGRI. Undated. Annuaire de Statistique
Agricoles 1 979-1980. MINAGRI, Yaounde.

Service de la Statistique Agicole/MINAGRI. Undated. Annuaire de Statistique
Agricoles 1980-1981. MINAGRI, Yaounde.

Service Provincial des Etudes et Statistiques Agicoles de l’Extréme-Nord. 1986.
Rapport Annuel 1985-1986. SPESAEN, Maroua.

159

Service Provincial des Etudes et Statistiques Agicoles de l’Extreme-Nord. 1986.
Annuaire Provincial des Statistiques Agricoles 1985-1986. SPESAEN, Maroua.

Service Provincial des Etudes et Statistiques Agicoles de l’Extreme-Nord. 1988.
Annuaire des Statistiques Agricoles de la Province de l’Extreme-Nord 1986-1987.
SPESAEN, Maroua.

Service Provincial des Etudes et Statistiques Agicoles de l’Extréme-Nord. 1990.
Annuaire des Statistiques Agricoles: Campagne Agricole 1988/1989. SPESAEN,
Maroua.

Service Provincial des Etudes et Statistiques Agicoles de l’Extréme-Nord. 1991.
Annuaire des Statistiques Agricoles: Exercice 1989/1990. SPESAEN, Maroua.

Service Provincial des Enquetes Agoéconomique et de la Planification Agicole
(SPEAPA). 1991. Annuaire des Statistiques Agricoles 1980/1981 - 1989/1990,
Province du Nora. SPEAPA, Garoua.

Service Provinciale de l’Agiculture del’Extreme-Nord1989. Rapport Annuel. Campagne
1988/1989. SPAEN, Maroua.

Service Provinciale de l’Agiculture de l’Extreme-Nord. 1990. Rapport Annuel d'Activités:
Exercice 1989/1990. SPAEN, Maroua.

SODECOTON. 1977. Corrrpte Rendu Annuel, Campagne 76/77. Garoua.
SODECOTON. 1978. Compte Rendu Annuel, Campagne 1977/1978. Garoua.
SODECOTON. 1979. Compte Rendu Annuel, Campagne 1978/1979. Garoua.
SODECOTON. 1980. Compte Rendu Annuel, Campagne 1979/1980. Garoua.
SODECOTON. 1982. Compte Rendu Annuel, Campagne 1981/1982. Garoua.
SODECOTON. 1983. Compte Rendu Annuel, Campagne 1982/1983. Garoua.
SODECOTON. 1984. Compte Rendu Annuel, Campagne 1983/1984. Garoua.
SODECOTON. 1985. Rapport Annuel, Campagne 1984/1985. Garoua.
SODECOTON. 1986. Rapport Annuel, Campagne 1985/86. Garoua.
SODECOTON. 1987. Rapport Annuel, Campagne 1986/1987. Garoua.
SODECOTON. 1989. Rapport Annuel, Campagne 1988/89. Garoua.

SODECOTON. 1990. Rapport Annuel, Campagne 1989/90. Garoua.

160

SODECOTON. 1985, 1986, 1988, 1989, 1990. "Fiches Techniques" for various
commodities. Garoua.

SODECOTON and Republic of Cameroon. 1989. Performance Contract, pp. 4-6.
SODECOTON. 1990. La SODECOTON: Note de Presentation. Garoua.

SODECOTON. 1990. Consommation des Intrants, Decompte des F orfaits, Rapprochement
Coats/Intrants. Garoua.

Ta’Ama, Mofi and Chevalier Endondo. Undated (1985). IRA-Bean Cowpea CRSP, D
Niébé, Rapport Partiel Campagne 1985-1986, Reunion IRA-SODECOTON. IRA, l.
Maroua.

Ta’Ama, Moffi and Chevalier Endondo. Undated (1985). IRA-Bean Cowpea CRSP,
Niébé, Réunion Programme Legumineuse IRA, Maroua.

Ta’Ama, Moffi. Undated (1986). Highlights of 5 Years of Cowpea Research in Cameroon,
Bilan de 5 Annees de Recherche sur le Ni6bé au Cameroun, IRA-an Cowpea
CRSP. IRA, Maroua and USAID, Yaounde.

 

 

Ta’Ama, Moffi. 1987. Programme National Légwnineuses, synthese des Resultats de
Recherches 1987. IRA, Yaounde.

Ta’Ama, Moffi, and G. Ntoukam. 1988. Bean Cowpea CRSP/IRA/USAID, Annual
Report 1988. IRA, Maroua.

Ta’Ama, Mofl'i. 1989. Short Final Report, Notes and Recommendations for Coupea
Research in Cameroon and 1989 Annual Report. IRA, Maroua.

USAID/ Cameroon and United Republic of Cameroon. 1979. Project Grant Agreement
Between the United Republic of Cameroon and the United States of America for
National Cereals Research and Extension USAID/ Cameroon, Yaounde.

van de Walle, Nicolas. 1989. "Rice Politics in Cameroon: State Commitment,
Capability, and Urban Bias." The Journal of Modern African Studies, 27:4:579-
599.

van de Walle, Nicolas. (Forthcoming). "The Politics of Non-Reform in Cameroon."
' ~ - Ed. by T. M. Callaghy

 

and J. Ravenhill. Columbia University Press, New York.

Wolfson, Jane. 1988. Bean/Coupea CRSP: Preliminary Assessment of Cowpea Storage
Methodologies Used by Small Farmers in Northern Cameroon. Bean / Cowpea
CRSP Project Report, Department of Entomology, Purdue University, West
Lafayette, IN.

 

161

Wolfson, Jane. 1991. Bean/Cowpea CRSP: Further Analysis of Coupea Production and
Storage Methodologies Used by Small Farmers in Northern Cameroon.
Bean / Cowpea CRSP Project Report, Department of Entomology, Purdue
University, West Lafayette, IN.

World Bank. 1980. Cameroon: Northern Province Rural Development Project, Project
Appraisal Report. World Bank official document, 1980.

World Bank. 1987. Cameroon Country Economic Memorandum. Report No. 6395-CM,
Western Africa Regional Ofiice, World Bank, Washington, DC.

World Bank. 1989. Cameroon Agricultural Sector Report. Report No. 7486-CAM,
Volume 1: Main Report. World Bank, Washington, DC.

World Bank. 1989. Cameroon Agricultural Sector Report. Report N o. 7486-CAM,
Volume 2: Statistical Volume. World Bank, Washington, DC.

World Bank. 1990. Project Completion Report, Cameroon, Northern Province Rural
Development Project (Credit 1075-CM/Loan 1919-CM). Report No. 8420. Africa
Regional Office, World Bank, Washington DC.

"Iilllllllllllllllli