_ 2
111...!
:.

kahuna”.

. , . fi mt.

., h . we“?
‘

..

‘ z . —
tit-Iurbri

. . ‘3. .
st . T. x
1L... 1 p 2. am -

:‘v. v.1. . ..

l.
.
..

.5 ,m...s._.~.i «\
7-,.- . V

- £5- 3
31% J: «3
$1} a

,. A.
til”. 19-
»Wxfi.‘ ‘33.
53349,
gal. ~P
. .|.

:s .51. v.3
.31.: 14 p.

.. .
v L 3:53.!
:2 .15?!

t.

 

\-\.o

2.7. In.

l.‘ 11‘
.\,

,.§. .. .5

“.5an4 .

V . . uluu¢ .
I. E s . -w'kloV, [ 1“ ,. 5,.
o 1. l (ctr V,“ V . I .2: . . 1." a ALF-.1...“ 1.9

! I" 'l'

.. «<2. .0! .
‘ .2... 3 .. 5,.

I‘ll!

 

A. P201.
‘1.’ \l in; 03“

Sal .
1103:!

71.32:.

.A . rich... I. (a.

it‘ll»...
.l)?

'1

. ELVNVLJU?!“ .

I

ESE... I

 

Op".
1:...

at ‘
.35 {I’I

3 1293 00769

llllllllll'llllllllll||1|| llll lllll Illxllllll L

 

""9 LIBRARY
Mlchlgan State
; University

 

 

This is to certify that the

dissertation entitled

MARKET REFORMS, FOOD SECURITY AND THE
CASH CROP-FOOD CROP DEBATE IN
SOUTHEASTERN SENEGAL

presented by

Stephan J. Goetz

has been accepted towards fulfillment
of the requirements for

Ph.D. . Agricultural Economics
degree m

WM

Major professor

 

 

 

May 7th, 1990
I)ate

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

uovum I MR 1 5 1995

._.'. ‘9 -..__. . . ..

 

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_______J _
fire—If?

MSU I: An Affirmative ActloNEqual Opportunity Institution
empmS-pd

 

MARKET REFORMS, FOOD SECURITY, AND THE CASH CROP-FOOD CROP
DEBATE IN SOUTHEASTERN SENEGAL

by
Stephan J. Goetz

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Agricultural Economics

1990

l a \K

I i\ NT 2\.\

év46— 0768;;

ABSTRACT

MARKET REFORMS, FOOD SECURITY, AND THE CASH CROP-FOOD CROP
DEBATE IN SOUTHEASTERN SENEGAL

by

Stephan J. Goetz

The Government of Senegal has set the goal of achieving 80% food self-
sufficiency. Policy instruments initially chosen to attain this goal include a producer
floor price for cereals; increased sales on credit of cereals fertilizer and improved seed
varieties; and reduced sales on credit of peanut seed. The general intent of these
policies is to encourage farmers to switch from eash to food crop production. In order
to inform decision-makers about the cost-effectiveness and likely effects of these poli-
cies on household food security, questionnaires were administered to 215 farm house-
holds located in the high rainfall, southeastern part of Senegal. Surveys covered
opinions of household heads; manifest production and marketing behavior; and potential
behavior in the form of responses to hypothetical questions. Tabular and econometric
analyses were carried out at the individual household member and at the aggregate
household level.

Five principal conclusions emerge regarding the policy instruments chosen by
policy makers. First, while private agricultural credit is indeed a severe constraint,
making draft equipment more available, rather than fertilizer, would have higher payoffs,
especially in the Casamance. Second, there is an important complementarity between
peanut seed used to attract additional workers to the household and the increased food
production made possible with those workers. Third, economies of scope in producing
both food and cash crops on the same farm leads to better utilization of resources. This

finding complements the previous one in suggesting that the cash crop-food crop trade-

ofl' in southeastern Senegal is less severe than commonly imagined. Fourth, farmers
prefer a variety of cereals in their diets, and this needs to be considered along with
relative processing costs and storability of different cereals in designing national food
policies. Finally, many households in southeastern Senegal are currently unable to
produce enough food to meet annual consumption needs, and in the short run may have
difficulty responding to a floor price incentive. Attention needs to be focused on raising
the productivity of rural labor working both on and off the farm.

ACKNOWLEDGEMENTS

I gratefully acknowledge the contributions of all those who have made this
dissertation possible. My foremost thanks are due to Michael T. Weber for his unfailing
support throughout my graduate training. including the field work in Senegal and the
completion of this thesis. The other members of my committee, Eric W. Crawford,
James D. Shaffer and John M. Staatz, provided research support in Senegal and on
campus in the preparation of the thesis. In addition, John S. Holtzman on many occa-
sions shared his experience, and James F. Oehmke commented on portions of this thesis.

My appreciation also goes to former colleagues Jacques Faye, P. Leopold Sarr,
Ismael S. Ouedraogo, Bocar N. Diagana, A. Abdoulaye Fall, Bashir Diop and, especially,
Ousseynou Ndoye. Janet Munn, Chris Defouw, Sally Petersen, Alioun Dieng, Fatoumata
Mbengue, Margaret Beaver and Mamadou Mane provided important services to the
project. The surveys were implemented by Abdou Karim Diallo and Papa D. Diack
(supervisors) and Godel Ba, Ousmane Sakho, Djibril Diop, Souleyman Balde and
Youssef Camara (enumerators). Results presented here would of course not have been
possible without the collaboration of the farmers, traders, and chiefs of farmer
organizations surveyed.

The study was funded jointly by the Bureau of Science and Technology, the
Africa Bureau, USAID Washington, DC, and USAID, Dakar, Senegal. In addition, the
overall study benefitted from discussion with Wayne N ilsetuen, Moribijan Keita, Lamine
Thiam, John Balis and Richard Caldwell.

Finally, for contributing more than they can imagine, my deep appreciation goes

to my family, near, far, and not quite so far away.

iv

TABLE OF CONTENTS

PAGE
LIST OF TABLES ............................................... vii
LIST OF MAPS ................................................. x
LIST OF FIGURES ............................................... xi
ABBREVIATIONS ............................................... xii
CHAPTER
I. INTRODUCTION AND RESEARCH CONTEXT .................... 1
1.1. A Brief History of Agricultural Policy in Senegal ................ 2
1.2. The New Agricultural Policy of 1984 ......................... 3
1.3. Objectives of the Study ................................... 7
1.4. Data Availability in Senegal ................................ 10
1.5. Sampling and Data Collection .............................. 11
1.6. Organization of the Thesis ................................. 14
H. FARM-HOUSEHOLD BEHAVIORAL CONCEPTS .................. 18
2.1. Food Security and the Scope for Price Policy ................... 18
2.1.1. Short Run Efiects of Food Price Changes .............. 19
[i] Production: Pre-Trade .......................... 21
[ii] Consumption: Post-Trade ........................ 21
2.1.2. Intermediate Run Adjustments ...................... 27
2.1.3. Food Price Policies in the Long Run .................. 31
2.2. E—S-P: A Dynamic View of Household Behavior ................ 34
2.3. Rural Organization: Farm Households as Coalitions .............. 47
2.4. Summary and Conclusion ................................. 62
III. HOUSEHOLD ORGANIZATION AND THE USE OF INPUTS ......... 64
3.1. Resource Inventories ..................................... 64
3.1.1. Structural Organization of Households and Labor Supply . . 64
3.1.2. Draft Equipment and Animals ....................... 69
3.1.3. Chemical Products ................................ 71
3.1.4. Seeds ......................................... 72
3.1.5. Factors Associated with Equipment Ownership .......... 74
3.1.6. Private vs. Public Involvement in Input Markets .......... 80
3.2. Investment Priorities, Expectations and the New Input
Distribution Policy .................................... 82
3.2.1. Investment Priorities at the Beginning of
the 1987 Season ................................ 83
3.2.2. Expectations for Returns on Crop Investments ........... 86
3.2.3. Farmer’s Knowledge and Perceptions of Fertilizer ........ 88
3.2.4. Fertilizer Benefit-Cost Ratios Based on
Farmers’ Expectations ............................ 93
3.2.5. Willingness to Use Selected Seeds and Fertilizer ......... 97
3.2.6. Issues Involved in Privatizing Input Distribution .......... 101

CHAPTER III Continued PAGE

3.3. Interdependencies among Inputs ............................ 103
3.3.1. Model Specification ............................... 103
3.3.2. Results and Discussion ............................ 106
3.4. Summary and Conclusion ................................. 112
IV. HOUSEHOLD PRODUCITON BEHAVIOR AND POSSIBILITIES ...... 114
4.1. Crop Production Goals, Strategies and Recent Dynamics .......... 1 14
4.1.1. Crop Mixes and Production Levels .................... 114
4.1.2. Perceived Constraints to Extensification ................ 125
4.1.3. Considerations of Cereals Output Response

under the APS .................................. 128
4.1.4. Prospects for Expanded Maize Production ............. 131

4.1.5. Recent Crop Mix Changes, Reasons Given by Farmers and
Evidence Relating to Farmer Decision-Making .......... 136
4.2. Auxiliary Income-Generating Strategies ....................... 139
4.2.1. Livestock ...................................... 139
4.2.2. Off-Farm Income ................................ 140
4.3. Cost Function Estimation ................................. 143
4.3.1. Introduction .................................... 143
4.3.2. Cost Measures for Multiproduct Firms ................. 145
4.3.3. The Translog Cost Function ......................... 146
4.3.4. Estimation, Results and Discussion ................... 147
4.4. Summary and Conclusion ................................. 150
V. HOUSEHOLD MARKETING BEHAVIOR ........................ 152
5.1. Transactions Behavior of Households ......................... 152
5.1.1. Households’ Coarse Grain Cash Market Position ......... 153
5.1.2. Food Availability Before and After Transfers; Stocks ...... 158

5.1.3. Cash Crop Sales, Livestock Sales and Off-Farm Activities . . 168
5.1.4. Consumption Preferences; Storage and Processing Issues . . . 174

[i] Unconstrained Cereals Choice .................... 174
[ii] Processing, Firewood and Storage Issues ............ 177
5.1.5. The Special Role of Rice ........................... 178
5.1.6. Opinions of Market Reform ........................ 179
[i] Opinions of Select Cereals Policy Measures .......... 180

[ii] Private Traders and Farmer Organizations:
Which Mix? ................................. 185

[iii] Marketing Responses Under Doubled

Cereals Production ........................... 187
5.2. Marketing Equations for Coarse Grains and Rice ............... 190
5.2.1. Recent Estimation Efforts .......................... 190
5.2.2. Model Specifieation Under Uncertainty ................ 192
5.2.3. Model Specification with Transactions Costs ............ 195
5.2.4. Empirical Results ................................ 200
5.2.5. The Rice Regression Equation ....................... 205
5.2.6. Policy Implieations ................................ 207
5.3. Summary and Conclusion ................................. 208

vi

CHAPTER PAGE
VI. IMPLICATIONS FOR PUBLIC POLICY AND FUTURE RESEARCH . . . 209

6.1. Summary and Implications ................................. 210
6.1.1. Agricultural Input Distribution Policy .................. 210
[i] Credit as a Constraint .......................... 210
[ii] Agricultural Equipment Supply and Maintenance ...... 211
[iii] Fertilizer .................................... 212
[iv] Summary ................................... 213
6.1.2. National Crop Research Priorities .................... 214
[i] Labor-Peanut Seed Linkage ..................... 214
[ii] Economies of Scope ........................... 215
[iii] Sectoral Complementarities (Infrastructure) .......... 215
[iv] Food-Cash Crop Marketing ..................... 216
[v] Summary ................................... 216
6.1.3. Intra-Cereals Substitution .......................... 217
[i] Maize Production Constraints .................... 217
[ii] Processing and Transformation Issues .............. 218
[iii] Consumption Preferences and Tastes ............... 218
[iv] Summary ................................... 219
6.1.4. Output Market Reform ............................ 219
[i] Limits to a Floor Price Policy .................... 220
[ii] Alternatives to Direct Price Policies ................ 221
[iii] Beyond Crop Policies .......................... 222
6.1.5. The Privatization Debate ........................... 222
6.1.6. Privatization and the Interaction Among Technology,

Prices and Institutions ............................ 225
6.2. Limitations of the Study and Directions for Future Research ....... 227

6.3. Knowledge, Perceptions and Beliefs: Implieations for Extension
and Further Research on the Market Reform Process .......... 230
APPENDIX A-l: EQUIPMENT AND DRAFT ANIMAL PRICES ........... 232
APPENDIX A-2: PARTIAL BUDGET FOR HERBICIDE USE ............. 233
APPENDIX A-3: FERTILIZER BENEFIT-COST RATIOS ................ 234

APPENDIX A-4: PRINCIPAL FIELD MAIZE PRODUCTION PROBLEMS . . . 238

APPENDIX A-S: PRICES UNDERLYING COST DATA .................. 239
APPENDIX A-6: NOTE ON THE 'FORCED SALES” HYPOTHESIS ........ 240
APPENDIX A-7: RESERVATION PRICES FOR SELLING CEREALS ....... 241
BIBLIOGRAPHY ................................................ 242

LIST OF TABLES

TABLE PAGE
1-1 Policy Objectives, Instruments and Researchable Hypotheses ........... 6
1—2 F arm-Level Survey Instruments ................................ 15

2-1 Assumptions, Data Needs, and Classification of Households to Evaluate

Consequences of a Food Price Increase .......................... 33
2-2 Strategies for Achieving Food Security in Southeastern Senegal ......... 39
2-3 Objectives, Rights and Obligations in the Wolof Household / Coalition . . . . 59
2-4 Allocation of Costs and Benefits in the Coalition ................... 62
3-1 Number of Exploitation: per Caucasian ......................... 65
3-2 Household Labor Force Distribution ............................. 66
3-3 Average Labor Force Structure ................................ 67
3-4 Dependency Status and Kilograms of Seed Used; by Region ........... 69
3-5 Draft Equipment and Animal Ownership ......................... 70
3-6 Use of Chemical Products ..................................... 71
3-7 Dependency Status and Chemieal Product Usage .................... 72
3-8 Household Seed Usage ....................................... 73
3-9 Cash Crop Seeds (Kgs) as Labor Incentives ....................... 75
3-10 Variables Associated with Equipment Ownership ................... 76
3-11 Equipment Status as a Function of Parastatal Afl'iliation
of the Household ........................................... 79
3- 12 Equipment Borrowing and Size of Labor Force According to
Household Organization and Region ............................ 81
3-13 Farmer Priority Uses of 15,000 FCFA; Beginning of 1987 Season ........ 84
3- 14 Purchase Intentions for Farmers not Having Received Fertilizer
as of June 1987, by Region .................................... 85
3- 15 FCFA Returns Expected by Farmers for Two Investments in an
Agricultural Crop Activity .................................... 87

3-16 Responses to the Question 'Do You Believe You Know the
Quantitative Fertilizer Needs of Your Cereals Fields Better than an

Extension Agent?” .......................................... 88
3-17 Farmers’ Perceptions about the Relative Effieacy of Manure and Mineral

Fertilizer Applied to Maize .................................... 90
3-18 Reasons Given by Farmers for Preferring Organic or Mineral Fertilizer,

Applied to Maize ........................................... 92
3-19 Farmers’ Fertilizer Yield Response Perceptions (100 =- Base Yield without

Fertilizer and 1987 Rainfall) ................................... 93
3-20 Sensitivity of Benefit-Cost Ratios for Fertilizer Use on Maize

to Alternative Assumptions about Input-Output Prices ................ 94
3-21 Improved Input Prices "Acceptable” to Farmers ..................... 98
3-22 Recent Official Fertilizer, Peanut and Cereals Prices ................. 99
3-23 Labor Category Size and Efficiency Measures ...................... 107
3-24 Coefficient Estimates for Factors Affecting the Quantity of

Coarse Grains Seeded (kgs) .................................... 109
3-25 ML Coefficient Estimates for Factors Affecting the Use of Fertilizer ..... 111

viii

TABLE PAGE

4-1
4-2
4-3
4.4

4-5
4-6
4-7

4-8
4-9
4-10
4-11
4-12

4-13
4-14
4-15
4-16
4-17
4-18
4-19

5-1
5-2

5-3
5-4
5-5

5-6
5-7

5-8
5-9
5-10
5-11
5-12

5-13
5-14
5- 15
5-16

5-17
5-18

Mean Annual Household Crop Production, by Study Region, 1986 ....... 117
Annual Crop Production per AEP and Equipment Use ............... 119
Variables Associated with per Worker Production ................... 121
Production Variables Associated with Households’ Food Production

Sufficiency Status . . ....................................... 124
Primary Constraints to Expanding Area Cultivated in 1986 ............ 125
Hectares Farmers would Reportedly Cultivate to Improved Cereals ...... 129
How Farmers would Change Total Area Seeded to Cereals,

Concurrent with Receipt of Improved Seeds Under the APS ........... 130
Reasons for and Against Growing Field Maize ...................... 133
Factors Preventing Expansion of Field Maize Crops ................. 134
Willingness to Substitute Millet/ Sorghum for Maize ................. 135
Absolute and Relative Crop Mix Changes in 1987 over 1986 ........... 136
Household Level Stocks of Cereals and Peanuts, Stratified by

First Investment Preference for 15,000 F CFA ...................... 139
Ownership of Livestock in the Study Areas ........................ 140
Timing of the Off-Farm Activity ................................ 141
Off-Farm Activity Frequency Counts ............................. 142
Beginning of Off-Farm Activity .............................. ’. . . 142
Cost Function Statistics for Agricultural Households ................. 148
01.8 Coefficient Estimates for the Translog Cost Function ............. 149
Cost Measures for Cash and Food Crop Production .................. 150
Coarse Grain Market Participation Profile ......................... 153
Farm Household Characteristies and Coarse Grain Cash

Market Position: 1986/7; 10 Months ............................. 155
Relationship Between Food Production Sufficiency Status

and Net Coarse Grains Market Position .......................... 159
Household Cereals Transactions, Related to Region and

Equipment Use, 1986/7 (IO-months) ............................. 160

Household Marketing and Estimated Consumption Characteristics (per
AEC) Associated with Households’ Food Production Sufficiency Status . . . 163

First Reason for Crop Sales, 1986/7; 10 Months .................... 164
Incidence of Taxes on Households with Difi'erent Food

Production Sufficiency Status and Market Position ................... 165
Livestock Transactions and Food Production Sufficiency Status ......... 169
Livestock Transactions and Coarse Grain Market Participation ......... 171
Per Capita Net Livestock Transactions by Household Characteristic ...... 172
Off-Farm Income by Food Production Sufficiency Status .............. 172
Uses of Proceeds from Off-Farm Activities and Contribution to

Food Availability ............................................ 173
Unconstrained Choice of Cereals for Consumption .................. 175
Preferences among Cereals .................................... 177
Knowledge of the Cereals Floor Price ............................ 180
How Household Heads Say they would Change Areas Seeded

to Cereals under Different Official Price Policies .................... 181
Reasons Given For and Against a Cereals Floor Price ................ 182
Preferred Floor Price Level by Area Seeded to Cereals in 1987 ......... 181

TABLE PAGE

5-19 Market Access as a Constraint to Selling More Cereals ............... 183
5-20 Desirability of Free Cereals Trade ............................... 183
5-21 Reasons For and Against Free Cereals Trade ...................... 184
5-22 Preferred Source of Inputs Under the APS ........................ 186
5-23 Preferences for Organizing Sales of Cereals ........................ 187
5-24 Priority Uses of Proceeds from Sales of Surplus Cereals .............. 189
5-25 Probability of Participating in Coarse Grains Cash Markets
Versus Net Quantities Sold .................................... 201
5-26 Determinants of Household Behavior in the Coarse Grain Cash Market . . 203
5-27 Reduced-Form Estimates for Rice Purchases ....................... 206
LIST OF MAPS
MAP
1-1 Senegal .................................................. 12
1-2 Food Security Project Research Sites ............................. 17

LIST. OF FIGURES

PAGE
FIGURE
1-1 Cereals Supply Situation in Senegal: 1974-1986,
In Estimated Percent by Source ................................. 4
1-2 Prices of Major Primary Commodities, 1950-1987 (Actual)
and 1988-2000 (Projected) ..................................... 4
1-3 Task Calendar: Survey Activities, Senegal ......................... 13
1-4 Village Triad Sample Units .................................... 16
2-1 Household Cash and Food Production and Consumption Possibilities ..... 20
2-2 Household Production Data for Cash and Food Crops per Capita ....... 24
2-3 Implications of a Relative Increase in Food Prices ................... 29
2-4 The Recursive Entitlement-Strategy-Performance (ES-P) Continuum ..... 35
2—5 Details of the E-S-P Set of a Household in a Given Year .............. 37
2-6 Adjustments with Friction ..................................... 43
2-7 Cost Surface for Two Outputs .................................. 45
2-8 Vertical and Horizontal Extension of Households ................... 49
2-9 Dependency Status and Land Use ............................... 50
3-1 Crops Grown by Household Members ............................ 68
32 Time in the Crop Cycle at Which Farmers Would Apply Fertilizer ....... 89
3-3 Differences in Gross Benefit Cost Ratios Expected by
Farmers for Returns to Using Fertilizer on Maize under 1987
Rainfall Conditions and their Own 20,000 FCFA Cash Investment ....... 96
3-4 Predicted Fertilizer Adoption Rates at Alternative Fertilizer Prices ...... 100
3-5 When Farmers would Prefer to Purchase Fertilizer and Improved
Seeds, According to Mode of Payment ............................ 101
4-1 Crops Grown in the FSP Research Areas, by Northern and
Southern Region, 1986 ....................................... 115
4-2 CrOp Mixes in Senegal and the FSP Research Areas, 1986 ............. 115
4-3 Cumulative Frequency Distribution for per AEC Cereals
Production, by Area, 1986 ..................................... 118
4-4 Maize Production and Area Cultivated: SenegaL 1961-1986 ............ 132
5-1 Prices and Estimated Quantities of Coarse Grains Sold
per Market Day by Farmers in the Weekly Market of Ndiapto
(Arrondissement of Koungheul), October 1986-July, 1987 .............. 167
5-2 Producer Coarse Grain Prices in the Weekly Market of
Ndoga Babacar, Maka: August 1986-1987 .......................... 167
5-3 Estimated Quantities Of Coarse Grains Sold by Farmers ,
per Market Day in Ndoga Babacar, Maka: August 1986-1987 ........... 167
5-4 The Region of Autarky ....................................... 196
5-5 Probability Density Functions for Quantities Bought and Sold
Under Alternative Transactions Cost Assumptions ................... 197

F A0
FCFA

FSP
GOS
ISRA
MADIA
MDR
MSU
NAP
ONCAD
PPF
RDS
SAED

SODAGRI

ABBREVIATIONS

Association de Base des Producteurs (SODEFITEX)
Adult-equivalent-consumer

Adult-equivalent—producer

Agricultural Production Support Project (USAID, Dakar)
Bureau d’Analyses Macro-Economiques

Consumable Product Equivalents

Consultative Group for International Agricultural Research
Centre de Recherches Zootechnique

Commissariat a la Sécurité Alimentaire

Food and Agriculture Organization (United Nations)

Franc de la Communaute Francophone Africaine (West African Currency
Unit; 1 FCFA = 50 French Franes; 31 US. =- approx. 300 FCFA)

Food Security Project (ISRA/MSU)

Government of Senegal

Institut Sénégalais de Recherches Agricoles

Managing Agricultural Development in Africa (World Bank)
Ministere de Développement Rural

Michigan State University (USA)

New Agricultural Policy (Nouvelle Politique Agricole)

Office National de la Cooperation et d’Assistance pour le Developpement
Production Possibilities Frontier

Regional Development Society

Société d’Aménagement et d’Exploitation des Terres du Delta
Société de Developpement Agricole et Industriel

SODEFTTEX Société de Developpement des Fibres Textiles

SODEVA
SONACOS

USAID

Société de Developpement et de la Vulgarisation Agricole
Société Nationale de Commercialisation des Oléagineux du Sénégal
United States Agency for International Development

xii

CHAPTER I
INTRODUCTION AND RESEARCH CONTEXT

Like many African countries, Senegal is seeking to reform the political-economic
organization of its agricultural sector to reduce national dependence on world markets
for food. The perceived need for reforms has arisen out of Senegal’s colonial legacy
combined with recent changes in the world markets for primary commodities. A central
goal of the reforms planned in Senegal is to induce farmers to switch away from peanut
production and towards increased food crop production.

Two principal empirical issues implicit in this reform strategy form the central
focus of the present study. First, what is the precise nature of the relationship or "trade-
off' between food and cash crop production in Senegal? Are factors of production other
than land better utilized when cash and food crops are grown on the same farm? Are
there also sectoral benefits, for example in the creation of infrastructure, to growing both
types of crops in the same region? Second, which policy can be expected to be most
cost-effective and equitable in raising food crop production? Is a commodity-oriented
policy, such as a floor price for coarse grains, cost effective and likely to raise food
production? If an input distribution policy reform is envisioned, what type or types of
inputs should or should no longer be distributed?

The general goal of this study is to assist policy makers in addressing these two
key issues, using primary data collected in Senegal during 1986-87. The following section
of this chapter briefly reviews Senegal’s agricultural history to provide the context for
recent policy reforms in general and this study in particular. This is followed by a
synopsis of the New Agricultural Policy (NAP) formulated in 1984, and an expanded

2

statement about the objectives of the study. A discussion about the availability of data
in Senegal and the sampling and data collection methods used in the study follow, while
the last section describes the organization of the remaining chapters of the thesis.
1.1. A BRIEF HISTORY OF AGRICULTURAL POLICY IN SENEGAL

Since independence from French colonial rule in 1963, Senegal has by and large
continued the agricultural food strategy pursued since the early 1900’s: peanuts, the
principal cash crop, were exported to France while rice was imported from Asia for do-

1 Local coarse grains (millet, sorghum and maize) were consumed

mestic consumption.
primarily on the farms where they were produced, and marketed surpluses were limited.
Smallholder cotton cultivation was introduced into the higher rainfall, southeastern areas
of Senegal in 1963 as a cash crop alternative to peanuts (Lele, van de Walle and
Gbetibouo, 1989).

Perhaps the most salient feature of Senegalese agriculture was the involvement of
parastatals in virtually all aspects of draft equipment and cash crop input and output
marketing. Organized within rural peanut cooperatives, farmers received peanut seeds
and other inputs (fertilizer, fungicides) on credit during the planting season and repaid
(or failed to repay) production credits in the post-harvest season under alternative
reimbursement systems (Crawford et aL, 1987; Kelly 1988a). Other parastatals (such as
SAED in the Fleuve) operated with a mandate similar to that of the peanut parastatal.
Extension and marketing services, attached to the delivery of inputs on credit, were
provided through the parastatal system.

This organization of commodity assembly and purchasing services to agricultural
producers left little role for a private sector, except where special circumstances gave rise

to a parallel private market. These included the development of private sector rice

 

1Stomal-Weigel (1988), p.18, points out that at the turn of the last century peanuts were
not destined for sale (they were consumed on the farms where they were produced).

3
processing in the Fleuve, which drew heavily on technology and production infrastructure
created by SAED (Morris, 1988), and spatial arbitrage by small traders (or farmers)
where price diflerences across international borders created profitable opportunities.

In the late 1970’s, unfavorable events in international markets combined with
widespread drought to produce a severe balance of payments and public finance crisis,
shattering the public’s confidence in rural institutions. In 1980 the crisis led to the
closing of ONCAD, the largest parastatal, after it had accumulated sizeable debts.2
With this action virtually all formal-sector credit and equipment supply to principal
production areas were eliminated.

To a large extent the institutional sector and public finance crisis was precipitated
by declining cereals and peanut production mused largely by reduced rainfall, stagnant
or declining groundnut export revenues with poor long-term prospects, and a more than
tripling of rice imports from 100,000 tons in 1975 to 320,000 tons in 1986, with a high of
380,000 tons in 1983.3 Figure 1-1 shows only about 50% of national cereals needs are
estimated to have been produced in the country during the past decade. Figure 1-2
indicates that real prices for major primary commodities have tended to decline together
since 1950, although the real price of peanuts has declined by more than that of rice, so
that the terms of trade have moved unfavorably for Senegal.

1.2. THE NEW AGRICULTURAL POLICY OF 1984

Faced with a financial crisis in the late 19708 and dismal projections for the
cereals situation, the Government of Senegal (GOS) in 1984 formulated a new food
strategy. The primary goal is to achieve on average 80% cereals self-sufficiency by the
year 2000. Specific objectives of the new food strategy include increased cereals

 

2This was not the first time that large amounts of credit given to farmers were simply
forgiven.

3 Crawford, 1988.

 

 

FIGURE 1-1: CEREALS SUPPLY SITUATION IN SENEGAL:
1974-1986. IN ESTIMATED PERCENT BY SOURCE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80 Legend
Nat.Prdn.
7o«r
__.__ Imports
60+ F’d.AId
E
a so a.
g
I'- 40 4b
25
E 30 4»
K!
20 "-
10.. ‘ '
19'74 V 19.76 - 19'78 f 1950 19.82 19184 Y roves
YEAR
Source: 1’. Martin (1988)
FIGURE 1-2: PRICES OF MAJOR PRIMARY COMMODITIES
1950-1987 (ACTUAL) AND 1988-2000 (PROJECTED)
2280 [ LEGEND
20254~/\ ll _- com on
Areoo-I- . 1“ Rice
§ 13754» 'A. ' ‘ \ '_ coin"
b \‘ j\/ ---- Mala.
; 1350+ ./\ I V \ _ Sorghum
3:: V . .
" g 1125‘
)-
3” soo<
£3.
:3 e7s<
‘15.
450‘
225‘
o r.-...rffir:e:ef. ‘ - A- * f-r.--fir.

1950 54 so ez so 70 74 7e 32 as so 94 so
52 so so 54 se 72 75 so 34 as 92 so 2000
YEAR
SOURCE: WORLD BANK (1988)

 

 

 

5
production through substitution of coarse grain for peanut production, combined with an
intensification and extensification of coarse grains production (FAO, 1984; USAID/APS,
1987). Maize has been singled out due to its allegedly profitable yield response to
mineral fertilizer. A second objective includes expanded regional trade in domestic
coarse grains and, a third, continued state withdrawal from input and output marketing
(MDR/GOS, 1984).

Since the GOS arrived at these national self-sufficiency objectives drawing upon
both economic and political considerations, the purpose of this study is not to calculate
an economic benefit-cost ratio for achieving national food self-sufficiency, a task
requiring more data than currently available. Instead, taking the broad policy environ-
ment and national objectives as predetermined in the short-run, the study attempts to
identify a combination of specific policies and programs that can reduce the cost of
achieving food self-sufficiency (this may be thought of as raising the benefit-cost ratio, or
at least maintaining that ratio, while paying attention to the resulting distribution of rural
incomes). Examples include aSsessing the anticipated consequences of reducing parasta-
tal activity to the benefit of private traders; the effect of emphasizing agronomic
research on cereals to the detriment of cash crops; and eliminating the distribution of
peanut seed to farmers on a credit basis. Making decisions to assist in achieving the
goal of greater food self-sufficiency requires a basic understanding of Senegal’s agricul-
ture, mainly at the micro- but also at the macro-level, to the extent that macro variables
affect micro-level behavior. It also requires a knowledge of technological relationships
and trade-offs, for example, between cash and food crop production at the farm level.

Given the GOS’s stated objectives, and the policy instruments chosen to achieve
each objective, a number of implicit and researchable hypotheses can be established
(Table 1-1). For example, to increase cereals production, a pan-seasonal and pan-

territorial producer floor price is to be implemented, presupposing a certain willingness

 

 

TABLE l-l:

POLICY OBJECTIVES,a INSTRUMENTS AND RESEARCHABLE HYPOTHESES
Policy Policy Researchable Thesis
Objective "Instrument" Hypothesis Chpts.
Increased Cereals Producer floor Producers willing V
Production & Sales. price for cereals. & able to respond
(Raise Rural Income?) to price incentive.
Emphasis on Maize.

Increased sales of Fertilizer & Seed III
cereals fertilizer profitable, and
and improved seeds. limiting factors
of production.
Reduced sales of Minimal economies IV
cash crop inputs of scope in cash and
(peanuts). food crop production.
No interdependencies 111
among labor & peanut
seed.
Expanded Trade in Remove legal barriers Existing rules V
Domestic Cereals. to cereals trade: constrain trade;
"Market Liberalization." potential markets
already exist.
State Withdrawal Credit for private Private Sector: III
from Input / Output cereals traders Waiting to "fill &
Marketing. (from banking system). in the vacuum”; V
Emphasis on farmer More efficient/
organizations: cost effective.

 

a. An additional objective is the expansion of (irrigated) rice and maize production

in the Fleuve.

"Market Privatization."

 

 

7
and ability of farmers to respond to output price incentives. Sales of fertilizer and
improved cereals seed to farmers presume that a lack of fertilizer and improved seeds
currently constrains cereals production.

Similarly, the removal of legal barriers to regional cereals trade (”market
liberalization") presumes existing regulations constrain trade in cereals, and that poten-
tial sources of demand and supply will readily be articulated once the legal rules change.
The provision of credit to private cereals traders (USAID/APS, 1987) assumes the pri-
vate sector is prepared to ”fill in the vacuum” left by state withdrawal fairly rapidly and,
more generally, that it can carry out input / output marketing functions more cost-effec-
tively in the long-run than parastatals.

The table also serves as a reminder that market liberalization is a question of
degree rather than a discrete choice between private versus public coordination of eco-
nomic activity.4 On the one hand, all trade in cereals within the country is to be "free",
while on the other hand, the government intends to influence price formation by manda-
ting and supporting a cereals floor price. A farmer’s freedom to sell at a floor price,
provided it is effectively implemented, becomes a private trader’s legislated constraint
when providing arbitrage services over time and between food deficit and surplus areas.

In retrospect, and as will be demonstrated in this study, the GOS was rushed into
choosing these policy instruments without adequate knowledge about the economic
reality of rural Senegal This was largely due to the poor rural data base in the country,
as discussed in Section 1.4.

1.3. OBJECTIVES OF THE STUDY
The specific objectives of this study are [1] to describe and analyze the existing

economic system facing farm households, and to explain the behavior of households

 

4See also Staatz et al. (1989) on this point.

8
towards input use and investments, production and marketing; [2] to evaluate the ability
of rural households to respond to the NAP and to assess the likely effects of NAP incen-
tives (intended or unintended) on the welfare of rural households in general and their
food security status in particular; and [3] to identify points of leverage in the broader
food system facing rural households that can complement current policy instruments in
achieving public objectives in an equally or more cost efiective manner, while at the
same time contributing to improved income distribution and economic growth in rural
areas of Senegal in the intermediate and long run.

While the primary focus of the study is on the behavior of rural households and
their ability to respond to the NAP, the analysis is extended to private traders where
relevant. In particular, the welfare of rural households under the NAP cannot be
adequately examined without understanding the anticipated efi'ects of transferring
input/ output marketing activities from parastatals to private agents.

The more specific research questions addressed in the study include:

1. Why are many households in southeastern Senegal poor and living at or
near subsistence? If there are better-off households or "success stories",
what can be learned from them that is transferable to the rest of the rural
population?

2. How does the household obtain access to factors of production and why
do some households use more resources than others? Why are rural
households organized the way they are; can complementarities between
labor, equipment and technical inputs be exploited to increase total
agricultural production; and what are the anticipated effects of privatizing
input distribution activities.

3. How do households decide on crop mixes to be grown and which variables

subject to policy influence affect household output and incomes? Are

9
there complementarities between cash and food crop production and
marketing?
Which variables affect the behavior of households in the coarse gains
eash market and how can these variables be influenced by policy so as to
increase regional surpluses of food gains marketed while also assuring
food security at the level of individual households?
Which strategies do households pursue to achieve food security and long-
term survival goals; how are these affected by public policy (reform), i.e.,
changes in the institutional environment?
What can be learned from farmers’ potential behavior (as indicated by
responses to hypothetical questions concerning investment priorities and
opinions about policy reform) to inform decision-makers in planning and
implementing cost-effective rural development policies that do in fact
achieve desired objectives?
Finally, can we develop an operational model of farm household behavior
that recogrizes individual agents’ cogtitive limitations under genuine
uncertainty and yet allows for generalizable policy prescriptions beyond a
description of agent-specific rules of thumb? Such a model should be
helpful in informing policy makers about market reforms/ structural
adjustment in other food deficit African countries (with at least similar
environments—cf. Eicher (1982) on the diversity of African countries),
rather than be limited to the geogaphic confines of this study. A starting
point for such a framework will lie in endogenizing constraints faced by
food system participants, i.e., in allowing the legal rules and institutional
"givens' determining the status quo to enter the analysis as variables so

that they may in turn be subject to policy design and influence. This last

10
question is not answered definitively. Its purpose is to guide the analysis
in the thesis.
1.4. DATA AVAILABILITY IN SENEGAL

policy changes and planning for the resumption of gowth in [African]

agiculture are hampered by a pervasive lack of country-specific informa-

tion. Reform efforts all too often try to apply general remedies to

Africa’s diverse problems.

U. Lele, 1989, p.45.

The agicultural data base in African countries is generally poor (Lele, 1989;
Eicher, 1982) and Senegal is not an exception. Aside from general census data collected
every 10 years or so, and official annual regional agicultural production data of ques-
tionable quality (Holtzman, 1987, discusses problems with using official data), there is
very little empirical information on rural households in Senegal; this makes it difficult to
construct even a simple food balance sheet for the country.5 When farm-level informa-
tion is available, it tends to be concentrated in certain parts of Senegal such as the
mam; Basin, which has historically attracted the predominant share of researchers,
notably French agonomists and farm management specialists.

While socio-economic household data are now being collected by Senegal’s
Agicultural Research Institute (ISRA), the data are from specific regions, such as the
Fleuve, Kaolack and Ziguinchor. Consequently, there was a large part of the country
where little systematic socio-economic information had been collected prior to the Food

6

Security Project surveys. Data collection was all the more urgent since it was precisely

 

5The construction of food balance sheets is advocated as a first step in food policy
analysis by Tirnmer, Falcon and Pearson (1983). These difi'iculties notwithstanding, Martin
(1988, pp.68-71) has constructed regional food balance sheets for Senegal using official data.

6At the time of these surveys, applied crop research was being initiated by the cotton
parastatal SODEFITEX (Cellules Recherches et Developpement and Suivi et Evaluation)
in Tambacounda, while research on livestock and pastoral systems was already underway
at ISRA/CR2 in Kolda.

11

in these areas, where rainfall tends to be high relative to the rest of the country, that the
government hopes to achieve its cereals self-sufficiency objective.

Select references providing information on southeastern Senegal, at various levels
of aggegation, include the AgoProgess Report (1970); Bertrand and Valenza (1970);
B. Diop (1986); Havard (1986a,b); Jolly, C. et a1. (1985); Landais (1985); MDR/SODE-
FITEX-SONED (1980); SODEFITEX (1986 Annual Report) and SONED-SODETEG
(1977). The information available in each of these references is reviewed in Goetz with
Dieng (1987). This report also presents a shift-share analysis for regional production of
principal crops for the period 1961-1986 using (official) secondary data. The analysis
reveals that the Casamance and Senegal Oriental (Tambacounda), despite their higher
annual rainfall, are losing out in their share of national millet/ sorghum and maize pro-
duction, while the Peanut Basin’s share is increasing. The following section describes the
data collection method used in this study, including the sample desigr and a table of
questionnaires administered to farm households.
1.5. SAMPLING AND DATA COLLECTION

Southeastern Senegal (see Map 1-1) was selected as the research area for two
reasons. First, this area has been relatively neglected in recent socioeconomic surveys
(see Section 1.4).7 Second, this is the principal area in which the government plans to
implement the New Agicultural Policy (MDR/GOS, 1984). The procedures used to
identify villages and households for the final sample are discussed in Goetz and Diagana
with Diallo (1987).

A research planning matrix and activity calendar (Figure 1-3), described in more
detail in Crawford et a1. (1988), were used to guide the survey work. Agicultural

households, defined from the production side (i.e., as units of production-see

 

7Hence the survey work was intended to spatially complement related research of the
macro-economic analysis unit (BAME) of Senegal’s Agicultural Research Institute (ISRA).

12

MAP l-l: SENEGAL

 

 

as emanco

 

13

FIGURE 1—3: TASK CALENDAR: SURVEY ACTIVITIES, SENEGAL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

YEAR 1986 . 1987

IOITR SEP OCT IOU DEC JAI PEI III APR INT JUI JUL ADC 58? OCT
mm I l I H II II I
ACTIVITIES I Harvesting I [Land prepll PtantingleredingjIflarvestj
MARKETING | l
ACTIVITIES | Marketing I Buy inputs

CLIMATE <-Rains-> < ----- Cool Uinds~----> <-----not winds ----- > < ------------- Rains ----------- >

Least 50.0: 5.9m

LOCAL REFERENCE

I

I

1

mm: Lobutoy i
assessors?

Lotti bu wow / Caron ISeebet Psx I Biir Newer [Lotti buy toy

 

 
    
   

DATA\HOITI

Secondary Date a
Review

 

Reconnaissance as:
Survey

 

Village/flu
Selection

 

Hire/Train 353
Enumerators

 

ill

Village Leader
interviews

 

Producer Census

Ell

 

Producer 1

Questionnaire:
Protest
interview . . -
0". En: W :2..:;::::::::

 

 

Producer ii
Questionnaire:
Protest ggg
interview gaggg
Data Entry ,m.3

Producer iii

Questionnaire:
Protest
interview
Data Entry

 

 

 

Producer xv

Questionnaire:
Pretest
interview ”,,,.
Data Entry Eggg

a. Benort-Catttn 8 -aye, 1552}

 

 

 

 

 

14
Benoit-Cattin and Faye, 1982) were chosen as basic units of observation in the farmer
surveys. While questionnaires (see Table 1-2) were administered to individual household
members when necessary (especially in survey 1b), most questions were addressed to
household heads.

The agicultural households were conceptually embedded within a food system
subsector (Shaffer, 1980), incorporating the market and trader, parastatal, and household
levels (Figure 1-4).8 The basic macro sample units were five triads, each consisting of a
weekly coarse gain market and two satellite villages, one of which had better and one

9 Each enumerator worked with a total of

which had worse access to the market village.
45 households in one triad. Two triads were located to the north of the Gambia
(”North”) while three were located in the Upper Casamance ("South"), as shown in Map
1-2. Overall study goals and resources available (2 supervisors and 5 enumerators)
resulted in this spatial organization of the survey sites.
1.6. ORGANIZATION OF THE THESIS

The remainder of this thesis is organized as follows. Chapter II provides con-
cepts for studying farm-household behavior and gowth under uncertainty. Chapters III,
IV and V examine input, production and marketing issues pertinent to the NAP,

respectively. Chapter VI presents a synthesis of the empirical findings, draws conclu-

sions and provides recommendations for agicultural policy makers in Senegal.

 

8F or a summary of results from the farmer organization- and trader-level surveys, see
Goetz et aL, "Executive Summary“ (1988), and the references cited therein.

9This stratification was intended to capture the effect of physical market access on
household marketing behavior.

 

15

TABLE 1-2: FARM-LEVEL SURVEY INSTRUMENTS

 

 

 

Survey Month
Number Contents Administered
Census Household head sex; ethnicity; number of active October
400 workers and total members; crop mixes; parastatal 1986
HHS participation; equipment ownership; land expansion;

storage out of 1985 production; off-farm incomes.
No. Is Detailed household demogaphy (listing of each November
215 member’s relationship to the household head, age 1986
HHS number of fields cultivated, worker status).

Literacy; recent changes in total area and reason;

crop mix of household head and recent changes;

livestock transactions; ranking of field size by

field owner; chemical input use.

N 0. 1b Seed and chemical input usage in 1985 & 1986 December
(member, seed/ input type, source, quantity used vs. 1986
desired, mode of acquisition, total cost, timeli-
ness, quality of inputs). Use of inputs by crop.

Detailed balance sheet of recent crop transactions.

No. 2 Equipment ownership (type, number, source, owned February
since when, condition) and use by crop. Parastatal 1987
participation (by parastatal), year of initiation,
why participate, satisfaction. Constraints to extensifi-
cation in 1986. Private credit transactions; livestock
transactions; crop production, transactions and stocks.

No. 3a Coarse gain transformation and storage issues; con- July
sumption preferences and changes (rice); peanut seed 1987
market opinion questions; non-agricultural activities;
chemical input acquisition (preliminary); livestock
transactions; crop transactions and stocks.

No. 3b Determinants of area seeded to cereals, opinions of August
cereals market reform; maize production problems; 1987
general questions related to fertilizer marketing.

No. 4a Investment preferences and opinions; fertilizer September
yield response; perceptions of and stated responses 1987
to the APS project.

No. 4b Opinions about the organization of the cereals market; October
peanut seed production constraints and anticipated 1987

responses to price changes; demogaphic changes; use of
fertilizer inputs (sources, quantity); livestock ownership.

 

 

 

 

16

FIGURE 1-4: VILLAGE TRIAD SAMPLE UNITS

 

DAKAR Retail/
& Cereals Deficit Hholesale
areas of Senegal

 

 

 

 

 

 

 

 

 

 

Regional Hholesale/
Centers Redistribution
flarkgt_§_lnadgr Rural Weekly Assembly
Lgygls Cereals Market
Village No. 1

 

 

    

  

10 HHS w/ RDS
10 HHS w/o RDS

Farm Lgvel \

Pa l v

 

    

  

Village No. 3
Poor access
to market

   

Village No. 2
Good access
to market

          

One
- Unit

 

10 HHS 10 HHS

       
   

 

Notes: HH - household
RDS - regional development society
5 triads * 43 households/triad - 215 hhs (maximum).

17

MAP 1-2: FOOD SECURITY PROJECT RESEARCH SITES

 

 

 
 
 
 
 
 
 
 
 
  

I Realm-1W
D Dogma»:
0 Warner
. Marketvtllage
— Nationairoaa
_— mammary
- ------ Ariana. boundary

I Wm...
"" Soumnudyuea

 

 

 

 

,\
TAIIIACOUNOA

I"??? ' .

    

 

 
 

ZIGUINCHOR
x"

 

 

CHAPTER II
FARM HOUSEHOLD BEHAVIORAL CONCEPTS

This chapter presents concepts of farm household behavior which are used to
guide the subsequent analyses in the thesis. Since African farmers are currently and on
average unable to feed themselves, let alone their nations, the concepts of food produc-
tion versus food consumption security in rural areas are examined in Section 2.1. More
specifically, the prospects for and potential of a food price policy, intended to raise food
production, is examined under different lengths of run. A preliminary analysis of 1986
per capita cash and food crop production data shows some households in southeastern
Senegal are unable to meet both their estimated cash and food needs out of crop
production alone. This motivates the discussion of an Environment-StrategieS-Per-
formance paradign in Section 2.2. Uncertainty and associated market imperfections are
included as key elements of the W facing food system participants. Also
included are consequences of these phenomena, such as market interdependencies,
credit limitations and transactions costs on the W adopted by farmers to survive.
Subsequently, W outcomes associated with the various strategies are examined.

A key strategy concerns horizontal and vertical extension or integation of
households. This is examined in Section 2.3. within the context of households as
coalitions of individuals with divergent economic interests. The argument is that
coalitions form in response to market imcertainty and high transactions costs; hence
households in a sense become a substitute for labor and food markets. The implications
of the concepts developed here for the subsequent analysis are summarized in Sec. 2.4.
2.1. FOOD SECURITY AND THE SCOPE FOR PRICE POLICY

The fact that Senegal imports an estimated 50% of national food needs, while
over 75% of the population is engaged in farming, implies a number of rural households

produce insufficient amounts of food to meet own-consumption needs. Food price policy

18

19

therefore must begin by examining whether or not rural households are food production
deficit or surplus, and what the characteristics of each goup are. In addition, the
amount of time allowed for farm-level adjustments to output price changes has to be
made explicit to fully assess the impact of a price policy through time. The amount of
empirical data required to predict the consequences of price policies evidently increases
with the length of the adjustment period. The purpose of this section is to provide a
framework for evaluating the consequences of and prospects for food price policies
under different lengths of run, and to specify the data needed to evaluate such policies.

2.1.1. Short Run Effects of Food Price Changes

To assess the distributional effect of an increase in the relative price of food in
the short-run, defined here as a period in which all farm-level resources are fixed

between alternative production activities, it is necessary to know food and cash crop

production levels per household consumer. In practice, the short-run refers to a period
between planting and the next year’s harvest, so that the household is not able to change
its output mix in response to the change in food prices. Under this scenario households
can be classified according to their pre-trade (production) and post-trade (consumption)

positions. 1

In a given gowing season, a household’s per capita output of cash (c) crops
and food (f) crops is given by a point in Figure 2-1. Line f‘ =I200 represents per capita
food requirements and q‘ refers to minimum annual cash needs per capita, consisting of
rural head taxes, clothing, and other necessary consumption items. Line (0,m) is a ray
from the origin through the intersection of q‘ and f‘. If food crop markets exist, a

household’s trading opportunities are given by a line with slope (-pf/pc), where poll,

 

1See also Reardon et al. (1988), who stratify their Burkinabé sample as food production
and/ or consumption secure.

20

FIGURE 2-1: HOUSEHOLD CASH AND FOOD PRODUCTION
AND CONSUMPTION POSSIBILITIES

 

 

 

 

 

 

C
Cash per m
capita a I IV
(FCFA) b a1. Cash
c \\ intensive
\
\\ FOOd
‘\\ intensive
Vrao
0‘ \
q* 31 \\\ \\\
K \\
b a \>\ \02
B \
2 ‘~ - / ) --
\‘K (Pf PC 0 Po
BO\\
\93
II 111 b a
0 F

f*-200 Food per

capita (kgs):f

Note: q* - minimum amount of cash per capita needed in the
household, 200 - fixed requirement of cereals (food). p -
price of cash crops, p - price of food crops (cereals).

Line '(pf/pc)o shows initial terms of trade between cash and
food.

21
through the point of cash and food crop production.2 Households can then be classified
as follows with respect to their pre- and post-trade positions.

[1] Production: Pro-trade. Households producing above line (0,m) in Figure 2-1
are W those below W producers, relative to cash
requirements q' and food consumption needs f‘. Further, if we consider only crop
production activities3 and q‘ as the minimum amount of cash required annually per
household consumer, the following is true prior to trade of households located in these
quadrants contained within the area given by lines OCmF:

Quadrant I : Food deficit, cash crop surplus producers;

II : Food and cash crop deficit producers;
III : Food surplus, cash crop deficit producers;
IV : Food and cash crop surplus producers.

Households located in quadrant II are producing insufficient levels of cash (from
cash crop production) and food crops to meet minimum subsistence needs of cash and
food. They rely on non-crop inflows—off-farm activities, transfers and gifts, etc.-to
survive, or they have reduced consumption levels.

[it] Consumption: Post-trade. We next examine how the situation of households
_ located in each quadrant changes if there are opportunities for trade. Households
producing in triangle Ia are food crop production deficit, cash crop production surplus,
and food and cash consumption secure: at prevailing prices a household located at al

can become food consumption secure by trading cash for food up to point “0 and remain

 

2Relative prices are assumed to be identical across all households for the sake of clarity.
In practice, buying and selling prices are likely to differ, which could be represented by
different slopes of the line (p0).

3The main concern here is with crop production activities since we are dealing with
allegedly agarian areas. If off-farm incomes and transfers are known, they can of course
be added to cash ”production” in Figure 2-1 since the y-axis measures (non-food) income.
These funds would result in an outward shift of the trading constraint.

22
cash surplus. A household in area Ib (e.g., at 81) can become food secure by trading to
80 but is forced into a cash deficit situation in the process. To pay taxes and meet other
required cash expenses, this household would have to draw on proceeds from non-crop
activities and external transfers to move from 80 to [q‘,200]. Alternatively, the house-
hold may be forced to forego food consumption security for the sake of meeting cash
expenditure requirements (i.e., move to a point to the left of f 3200). The difference
between households located in Ia and 1b, therefore, is that both are food production
deficit, but those located in Ia are able to become food consumption secure at relative
prices using only proceeds from agicultural activities, while those located in Ib have to
draw on non-crop incomes to become food consumption secure and continue to meet
cash expenditure requirements.

Households located in triangle 11a, for example at 82, can become food consump-
tion secure at prevailing relative prices (by trading to 80) but will become more cash
deficit in the process. They are thus in a situation similar to that of households in Ib.
Households located in IIb can not become food consumption secure at existing prices
using only cash proceeds from agicultural production. They rely on off-farm activities
and other transfers to meet both their food consumption needs (f‘) and cash expen-
diture requirements (q’).

Households in III are in a position opposite to that of households located in I.
For example, a household producing at B3 in IIIb can trade surplus food for ”cash" to
point 80. It would then also need off-farm income or transfers to move to [q',200].
This also shows that food production secure households are not necessarily food con-
sumption secure, since a household producing at 83 may have to give up food consump-
tion security to meet cash expenditure requirements. It can be seen, finally, that a
household located at (:2 is in the same position as a household at point al, except that it

trades in the opposite direction.

23

Rural head taxes reduce the amount of cash available to be traded for food by
shifting the trading possibilities line towards the origin. The higher the tax, the more
households producing in triangles Ib, Ila and 11b will have to draw on off-farm activities
and transfers to move towards [q‘,200], other factors equal. Similarly, it is possible that
households producing at a point beyond [q‘,200] become food consumption deficit
and / or fail to meet cash expenditure requirements as a result of the tax.

Using the concepts developed in Figure 2-1, a scatter plot of empirical data has
been created using information from households studied in the ISRA/MSU sample.
This helps show the potential distributional consequences of an increase in the relative
price of food in the short run. In particular, households producing insufficient amounts
of food to meet annual per capita requirements (f < 200) will be worse off with higher
food prices as measured by the reduced availability of cash under the deteriorated terms
of trade of cash for food, while households producing f > 200 will be better off. This is of
course not a surprising result, but it is significant to observe the large proportion of
households shown in Figure 2-2 which are worse off in the short-run with a higher price
of food.4

For lack of a better measure, it is assumed in Figure 2-2 that minimum annual
per consumer cash needs are given by the sample mean (q‘ ac-bar - 33,000 FCFA) of
the value of cash crops produced per capita.5 This is calculated as the value of peanut
(90 FCFA / kg) and cotton (100 FCFA/ kg) production divided by the number of adult-
equivalent consumers per household. The per capita food production sample mean, cal-

culated as the per adult-equivalent consumer quantity of millet, sorghum, maize and rice

 

4See Chapter III for a description of consumption requirement standards and the
conversion from unmilled to milled coarse gains.

5This compares favorably with a number of FCFA 30,000 per adult equivalent consumer
obtained in a survey of village chiefs for m cash and food expenditure needs of a typical
household.

CASH CROP VALUE/CAPI'I‘A IN FCFA

(Thousands)

24

FIGURE 2-2: HOUSEHOLD PRODUCTION DATA

160
150
140
130
l20
110
100
90
80
‘70
60
50
40
30
20
10
0

FOR CASH AND FOOD CROPS PER CAPITA

 

 

f-ZOOkg

 

T-srokg

  

Cl

  
  
  

K Expansion

Path (c-f)

-Ci---U---Ci---- --- --------------- -1 ------------------
Ci t. c-q'-33,000FCFA

 

 

0.4 0.6 0.8 1
(Thousands)
CEREALS PRODUCED PER CAPITA lN KGS

1.2

1.4

 

25
produced, is 310 kgs. The distribution of households among the different quadrants is as
follows (there is a significant difference between northern and southern regions):
I=5% II-32% III-=30% IV=33%

Consequently, 37% of the households would unambiguously be worse off under a regime
of higher food prices in the (very) short-run. Since higher cash prices would also affect
barter prices of cereals, households not participating in cash markets would also be
affected by the price increase. In Figure 2-2, a higher price of food is represented as a
clock-wise rotation of the trading constraints. From this it is possible to determine the
new number of households falling into the different categories after the price increase.6

With the exception of one household, the general trend for the production
expansion path in Figure 2-2 is towards cash (crop) intensive production (especially if
the line 0,m is redefined as the ray from the origin through the mean points of produc-
tion [33,000 FCFA, 310 kgs]). This may be due to the nature of the production surface
(cash-crop biased technology) and / or the higher expected profitability of cash crop pro-
duction, given the fixed prices of eash crops. It is noteworthy that the data in the
ISRA/MSU sample of households were collected in a year of adequate rainfall (1986),
following drought conditions in 1984/85. To some extent the average production of food
(310 kgs) may reflect the accumulation of security stocks (8) in anticipation of subse-
quent years of drought (i.e., E[s] - 110 kgs - 310-200 kgs).

Given information on transfers and non-crop farm income it is also possible to
specify the proportion of households among those producing f < 200 which would conti-

nue to meet food consumption and cash expenditure requirements after a food price

 

6Note on recent (1988) price policy changes in Senegal: Old and new prices for peanuts
are 90 and 70 FCFA/kg, respectively. For cereals they are unchanged at 70 FCFA/kg, while
for rice they have fallen from 160 to 130 F CFA/ kg. Data show a small increase in the price
of rice relative to peanuts (from 1.78 to 1.86), and a sizeable increase in the price of cereals
relative to that of peanuts (0.78 to 1.00). Food prices are official.

26
increase (i.e., only experience a reduction in q) and those which have to increase the
flow of off-farm incomes and / or transfers to meet these requirements. Similarly, the
number of households that are better off after the food price increase (those in quad-
rants III and IV), or can reduce their reliance on ofi-farm activities to meet cash
expenditure needs, can be determined.

This analysis shows how crucial the timing of the introduction of the higher food
price is for food production deficit households, assuming it in fact afiects all house-
holds.7 For households currently producing f < 200 which are able to increase food
production in response to higher food prices, whether the price change is introduced (or
announced) after or prior to planting can spell the difference between a season (a) of
food scarcity and (b) increased availability of crash combined with satisfactory food
consumption following a change in the output mix. In practice the government will not
deliberately mislead farmers in the timing of price changes, although public coffers have
to be protected against the possibility of exceptionally good harvests when floor prices
are fixed (example of two good consecutive cereals harvests in Mali). This may explain
why the official buying or floor price for cereals has in some years been announced
rather late in the growing season in SenegaL8

Households not able to respond to the food price increase will have few alterna-
tives to allocating more labor to non-crop production activities to meet cash and food
expenditure requirements. This brings us to intermediate run adjustments to food price

increases.

 

7See Chapter V.

8111 1988 the official peanut price reduction (to 70 FCFA/kg, i.e., corresponding to an
increase in the relative price of local coarse grains) was announced three months prior to
planting, allowing farmers to make adjustments.

27

2.1.2. Intermediate Run Adjustments

The intermediate run is defined as a period in which resources can be reallocated
between crop enterprises but are fixed at the farm-level (e.g., a two-year harvest-to-
harvest period, which constitutes the conventional short-run in supply studies) and from
crop to non-crop enterprises where that is feasible. In reality it may be fairly easy to
adjust the amount of non-family labor from one production season to the next. How-
ever, to the extent that hired labor is paid with food during the production season (it is
actually not possible to retain seasonal labor unless there is sufficient food in the house-
hold), food deficit households will suffer from a food price increase in two senses: (a)
by experiencing a higher cost of food to the family and (b) by a reduction in the use of
hired labor which in turn shrinks the production possibilities frontier (ppf) towards the
origin, other factors held constant.9

In the intermediate-run, with resources assumed to be fixed at the farm but not
the crop level, it is possible for the household to change its crop mix by producing at a
different point on its ppf. The size of the crop substitution will depend on the degree to
which inputs can be reallocated among cash and food crops, as measured by the slope of
the ppf, i.e., dc/df.10 A key variable determining the effect of a food price increase in
the intermediate run is the maximum quantity of food the household can produce per
capita with its fixed resources, i.e., the size of {max given c =- 0. If farmers maximize
profits and face identical expected prices, an upper bound for fmax is given by the inter-
section of the line of relative prices passing through the point of production (c0,fo) with

 

9It is important to distinguish seasonal workers, who eat regularly in the household
(sunny and W), from hourly wage laborers. See also Section 2.3.

10The elasticity of substitution among outputs (8c/8f)(f/c) depends on the degree to
which capital and labor are fixed in the production of each crop. In practice it may be easy
for the farmer to reallocate hoes, seeders and labor from cash to food crop production,
implying a fairly flat slope of the ppf leading to a (food) corner solution (assuming perfect
knowledge of prices).

28
the f-axis (due to an assumption of strict quasi-convexity of the ppf). The number of
households located in IIb (Figure 2-1) gives the minimum number of households not able
to produce f > 200 in the intermediate run. If the assumption of price homogeneity
across households does not hold and if data on price expectations in each household are
not available, it is necessary to estimate each household’s ppf given its fixed resources
(denoted by subscript 0), i.e., c a g(f,k0,lo) to determine fmax for c =- 0. This permits a
more precise, production-based classification of households according to the size of fmax‘

These results are summarized schematically in Figure 2-3. The upper diagram
shows a cash crop (c) and food crop (0 production possrbilities frontier (ppf) for
households capable of producing more than enough food to meet household consump-
tion requirements (ppfa). Households not able to meet household food needs, even if all
resources were devoted to food crop production, are represented by ppfb. The bottom
diagram traces out levels of household income as the relative price of food changes,
holding constant the intake of food at 200 kgs/eapita.

Three eases may be distinguished for a given set of relative prices: 1. households
producing f < 200 for which {max < 200 [i.e., f < 200 > fmax]; 2. households producing
f < 200 for which fmax > 200 [i.e., f < 200 < {max}; and 3. households producingf >
200 [i.e., f > 200 < {max} at current prices. The bottom portion of the figure shows
households in the first category are unambiguously worse off as the relative price of food
rises towards p2. An important implication is that unless these households can expand
their ppfs by reallocating resources from non-crop to crop activities, and / or through
input price and distribution policies, so that {max exceeds 200, they may be forced to exit
agriculture to work in the non-crOp sector. Given their current situation, these
households are unable to increase cereals production sufficiently to cover consumption

needs. Households whose members are unable to migrate to Dakar or rural towns

29

FIGURE 2-3: IMPLICATIONS OF A RELATIVE INCREASE IN FOOD PRICESa

    
  
 

 

 

 

 
    
  

 

 

 

C2
\ 130
q* __.--_----- .....
p1
po
c1 tin-III-c-CC-I-O-II-"C'
p: D:
f f
200
c|(23200)
prl
flux>200
ppfs: fuss (200
co r--------- a-----— -------
C: l— ----------------
qt _.__ —————————————
c1 l- ———————————————— .1
Pt
p _ _
PO D: D: Pc

a. Depending on whether or not the ppf is such that f>200
when c-O; intermediate run adjustments are assumed, with
resources fixed at the farm-level.

30
(and / or that are unable to send offspring) may go hungry to the extent that rural
demand for labor declines as the price of food rises.

Consequences for households in the second category [i.e., f < 200 < frnaxl’ as
measured by income q, depend on the size of the relative food price increase. In
particular, there is a range of prices in Figure 2-3 (from p0 to p2) in which income
declines, showing these households to be worse ofi' in the intermediate run. Households
in the third category, for which f > 200 under current prices, experience improved
incomes as they increase the production and sales of cereals in response to higher prices.

One important empirical task then is to estimate the proportion of households in
each of these categories. Targeted policy interventions in input markets can subse-
quently be used to increase the ppfs to fmax > 200, if that is deemed desirable, and
output price policies can then be used to influence the mix of cash and food crop
production without introducing adverse distributional consequences for households’
access to food.

To summarize, if expected output prices vary for individual households, it is
impossible to predict which households will be unable to respond to food price increases
until the shape of the ppf is known. In particular, when examining individual points in
cash crop-food crop space, it is not known a 12de how far each pfl' ”stretches" along the
food axis, nor how this intercept shifts outward if resources previously allocated to non-
crop production activities during the growing season are committed entirely to food
production at the new price of food.11 Given empirical data, it is, however, possible to

unambiguously estimate the number of households gaining and losing from an increase

 

11Households relying on off-farm income during the growing season to purchase food
to maintain labor productivity will expend more effort on such activities, further reducing
their agricultural ppfs. If perfect capital markets existed (through time), food production
deficit households could borrow funds against expected future revenues from sales of food
due to higher food prices and hence expand their ppfs.

31
in relative food prices, once the shapes of the different ppfs are estimated. Since the
data collected for this study do not include input allocations by crop, this analysis was
not performed.

2.1.3. FoodPricePolicleslntheLonann

Two principal research questions arise in predicting rural household responses to
changes in food price policies over the long-run, when household resources and produc-
tion technologies are variable. 12 The first concerns the rate and level of adoption of
existing technology (K) by individual households, which in traditional analyses depend on
the profitability of that technology via (effective) derived demand functions. Given
Senegal’s history of state intervention in agriculture, however, more than farm-level
demand equations for improved inputs are involved. For example, agricultural draft
equipment was historically made available mainly to the Peanut Basin, and not the
higher-potential agricultural areas of the Upper Casamance in Senegal (e.g., Havard,
1986a; Ndoye, 1980). Improved technology was therefore not necessarily distributed or
adopted according to expected economic pay-offs, but through political decisions which
in turn influenced the (ex-post) comparative advantage (i.e., factor endowments) of the
different regions of Senegal. To the extent that private markets for agricultural techno-
logy failed to develop, a broader examination and understanding of the public (parasta-
tal) sector in Senegal and its role in enabling the adoption, maintenance and ownership
of equipment through time is necessary (see also Sec. 2.2.).

A second question relates to changes in the relationship between inputs and
outputs over time, i.e., changes in the implicit production function M. In general, this
kind of analysis uses historical production and price data to evaluate whether techno-
logical change is biased towards a particular agricultural output or whether production

 

12The focus here is solely on technology, not labor.

32
functions are homothetic (see Antle, 1984; Kuroda, 1988; and the references contained
therein). Given the paucity of reliable time series data in Senegal it is impossible to
pursue this kind of analysis. Nevertheless, even though reliable research expenditure
data are difi'icult if not impossible to come by, and even though a recent World Bank
study (Lele et al., 1989) concludes more research is needed to increase the productivity
of cash crop production, it is probably correct to argue that most of the agronomic
research in Senegal and West Africa to date has been directed towards improving the
on-farm productivity of cash crops (see also the discussion in Eicher (1989, p.18)
regarding the low use of improved cereals varieties and the relatively large stock of on-
the-shelf technology for cash crops in Africa).

It can be shown that the ratio of cash to food crop production, at any given price
ratio, is higher the more cash crop-biased the technology employed. This also suggests
that the more food crop-biased the agricultural technology, the less food prices have to
rise relative to cash crop prices to achieve the same ratio of cash and food crop produc-
tion. However, if technological change is biased exclusively in the direction of food
crops and if prices of food eventually decline as supply meets demand, the theoretical
possibility of 'immiserizing farm-level growth" can not be ruled out.13

To conclude this section, Table 2-1 provides an overview of the assumptions, data
needs and classification of households to evaluate the consequences of a food price
increase in food deficit nations such as Senegal. It is argued that households in rural
Senegal have a variety of economic options for achieving food consumption security, and
that own-food production represents only one option. The argument is not that food
price policy is ineffective. Instead, the W of agricultural policies is critical, and
before food price policies can bring about broad-based distributional benefits it is

 

13The notion of immiserizing growth in the case of a country engaged in international
trade is discussed by Bhagwati (1982, orig. 1956).

TABLE 2-1: ASSUMPTIONS, DATA NEEDS, AND CLASSIFICATION OF
HOUSEHOLDS TO EVALUATE CONSEQUENCES OF A FOOD PRICE INCREASE

 

Length of Run

 

 

Assumptions Intermediate
or Data Short Run Run Long Run
Needs 12 months 12-24 mths 24 mths +
l-Resources Fixed within ‘ Fixed at the Variable
(K,L) crop enterprises Farm Level
la-Techno- Fixed Fixed Variable
logy [M]
2-Data Nbr. of consumers Units of labor Institutional
Needs Kgs food produced and capital used Environment
(cumula- Kgs cash produceg [output price] Input prices
tive) Exogenous income Investment
Output prices demand fctn.
Bias in M;
Heterotheticity
3-Notation c , f , a a m(c,f,k ,l )-0 m(c, f, k, l)-0
0 0 0 0 0
(a-ao) but k-g(xl )
and l-h(xl)
m-m(tim£)
4-Household 8q/5pf (given f (8c/8f)(6f/8p) Same as i/r
Response and c fixed) Totalk kl fixgd but k,l not
fixed
5-"Test" Percent of hhs Percent of hhs % of hhs w/
Statistic with f < 200 with fma x< 200 c/f - 1.0
6-Refined Cash/food f < 200 > ax
Classifi- production f < 200 < fmax
cation and consumption f >2 200 < fmax
secure or
insecure

 

a. This information is required if the intention is to distinguish
households becoming food consumption insecure from those which
merely experience a decline in cash consumption as a result of
the food price increase.

34
essential to determine households’ existing food production possibilities and (if neces-
sary) to expand these by relaxing non-price constraints and / or to seek ways of increasing
the productivity of labor in non-crop activities. This may be an obvious assertion, but it
is often ignored when well-meaning assumptions have to be used to compensate for a
lack of empirical data.

The following section examines the household’s decision problem in a more
dynamic and recursive perspective, drawing on concepts developed in the present
section, and on the work of Shaffer (1980), QB. Williamson (1986) and Sen (1987). The
remainder of this Chapter attempts to lay the conceptual groundwork for analyzing food
strategies pursued by households, taking into account the food system environment
facing farmers in southeastern Senegal, which are examined in the remainder of the
thesis.

2.2. E-S-P: A DYNAMIC VIEW OF HOUSEHOLD BEHAVIOR

Figure 2-4 shows the recursive relationship between select decision variables of a
household for two years, each of which is divided into two periods. Each year, input use
and crop mix decisions are made during period a (the rainy season), while marketing
and (net) investment activities take place in period b. Consumption is listed explicitly in
period b, even though it takes place throughout the year, to emphasize the maintained
hypothesis that consumption decisions are made at least in part after production is
known (i.e., at harvest). For example, an unexpected shortfall in cereals output due to
drought may force the household to change its mix and level of cereals consumption.

As in a multi-period recursive Linear Programming tableau, the performance of
one period feeds into the entitlements of subsequent periods (denoted as a transfer of
performance outcomes {Xij})’ thus shaping and constraining the choice set and, ulti-
mately, the performance of that subsequent period. The quantity of cereals produced in

period a of year n, for example, provides part of the entitlement to consume and sell

35

FIGURE 2-4:
THE RECURSIVE ENTITLEMENT-STRATECY-PERFORMANCE (ESP) CONTINUUN

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:(Year n-I)
. Year (n)
E Period as——q
S INPUTS
P PRODUCTION
" -Period b—-—-———
E MARKETING
S {X13} CONSUMPTION
i NET INVESTMENT
[-Year (n+1)
E -—Period a
{xij} S INPUTS
{xij} P PRODUCTION
l -—Period b
E MARKETING
{xij} S {xii} CONSUMPTION
{xij} i NET INVESTMENT

 

 

Year n+2 :

Note: As in the multiperiod or intermediate inputs linear programming
tableau, {Xi.} shows the transfer of performance outcomes (e.g.,
production, at investment in equipment, etc.) into the entitle-
ments of subsequent period(s).

Period a - rainy season.
Period b - dry season (post-harvest).

36
cereals in period b, but also provides the entitlement to cereals seed as an input in
period a of year n+1. .

Own-production provides only one side (i.e., the supply side) of the entitlement
to sell cereals; a complementary entitlement is the presence of a buyer willing to trade
cash for cereals at an acceptable price. The demand side of this entitlement may arise
in the form of a cereals harvest failure on a neighbor’s plot. Hence, neither the supply
nor the demand side of a potential exchange can be taken for granted (see also Shaffer
et aL, 1983; and Binswanger and McIntire, 1987).

The by and large random production outcome in year n therefore not only affects
the marketing, consumption and net investment opportunities or entitlements of that
year, but also the production possibilities in year n+1 via the quantity of cereals carried
over as seed stock and as food to feed the household during the crop growing season
(i.e., the hungry season). Not only resource endowments, but also the productivity of
resources, may change from year to year. For example, increased food availability in
period a will raise the productivity of labor, and fertilizer residues in fields previously
planted to cotton will tend to raise the productivity of that land in producing maize in
the second year. 14

Figure 2-5 shows in more detail the entitlement set, strategies and performance
dimensions of interest in this study. The diagram highlights how the performance
outcome of period a, along with variables beyond the immediate control of the house-
hold, such as access to parastatals and marketing opportunities, becomes the entitlement
set for the subsequent period (b).

Table 2-2 provides an overview of alternative strategies available to households in

south-eastern Senegal for achieving food security. Following Sen (1987), strategy A.3.

 

14Similarly, the nitrogen-fixing ability of peanuts (as a legume) benefits millet grown in
the same field in the following year.

37

FIGURE 2-5:
. DETAILS OF THE E-S-P SET OF A HOUSEHOLD IN A GIVEN YEAR

 

 

—Year n; Period a

 

Entitlement INPUTS

Set {E13} Labor
Equipment
Seed
Chemicals

 

1

Strategy (Sia} INPUT USAGE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CROP MIX
Performance {P.a} CROP OUTPUT
L— - Entitlemeni. CREDIT REPAYMENT —Period b—~
Set {51b} MARKET OPPORTUNITIES
Strategy {Sib} MARKETING
CONSUMPTION
NET INVESTMENT
T
Performance {Pib} CONSUMPTION ADEQUACY
INPUTS

 

 

 

 

 

 

 

 

 

 

Year n+1; Entitlement Sets, etc.

Note: i denotes the i-th household.
Some strategies (not shown) may be pursued throughout the year;
e.g.: - off-farm activities/incomes
- livestock raising and transactions
- borrowing credit for short-term consumptions purposes.

38
yields production-based entitlements to food. Strategies A4. and 3.1. involve trade-
based entitlements, while strategies 32. and B3. are based on transfer entitlements.
Different strategies involve different levels of complexity and riskiness. For example, to
trade cotton for cereals a seller requires a market outlet for cotton and at least one food
system participant willing to exchange cereals for cash at an acceptable price. More
complex strategies thus involve more interdependence and require higher degrees of
market coordination.

Reliance on food aid is risky since it often comes late, if at all. The borrowing
strategy, similarly, may fail if all households in a given area produce insufficient quanti-
ties of food, as is usually the case when entire regions of a country are stricken by
drought. Several combinations of strategies are possible, leading to more complex obser-
ved outcomes, and given the uncertainty associated with agricultural production, one may
expect to see households pursuing more than one strategy.

Non-agricultural activities are potentially important in that they can help house-
holds to survive crop failures and also to facilitate investments in agricultural inputs.

The steady income stream they provide may also increase the willingness and ability of
households to sell and buy food. Off-farm activities carried out during the growing
season (period a) may tend to conflict with agricultural production. Consequently, more
information is required to gauge whether non-agricultural activities "compete with” or
"complement" agricultural activities.

The choice of a given strategy would in principle depend on two key factors: (3)
its availability-i.e., whether or not the decision-maker is entitled to pursue it-and (b) its
expected effectiveness in contributing to the goal of achieving a minimum level of food
security and/ or cash income. This dichotomy is more apparent than real, however. For
example, if the cotton parastatal does not operate in a given region, the cost of inputs

needed to produce cotton becomes very high and the selling price of cotton very low for

39

TABLE 2.2:
STRATEGIES FOR ACHIEVING FOOD SECURITY
IN SOUTHEASTERN SENEGALa

 

 

A. CROP PRODUCTION-BASED STRATEGIES
l. Equipment-Use
0 use traditional hand tools
0 borrow entire set of draft equipment
a borrow partial set of draft equipment
a own a full complement of equipment
2. labor-Use Strategies
0 use family labor only
' use seasonal labor (sashes. means.)
0 hire hourly or daily wage labor
3. Food Crop Production (Self-Sufficiency)
o ' traditional coarse grains
(millet, sorghum and maize)
0 rice
4. Cash Crop Production
0 peanuts
o cotton

B. NON-CROP PRODUCITON STRATEGIES
1. Income From Other Sources
0 agricultural labor (off-the-farm)
o livestock
o non-agricultural activities
2. Gifts
0 traditional gifts (m Ma)
0 food aid
3. Borrowing
a cash

 

inputs such as herbicides, fertilizer, insecticides and
fungicides.

40

a farmer in that region, so that a strategy of growing cotton is ineffective in contributing
to food consumption security. The advantage of asking two (a and b) rather than one
question (only b) is that the institutional environment giving rise to different sets of
entitlements to produce and market cereals and / or cash crops becomes a variable in the
analysis, and thus subject to policy influence.

The concept of a demand function for certain factors of production (derived from
a production function under profit-maximizing behavior) has to be modified here to re-
flect the fact that observed outcomes-such as the use of herbicides-do not simply result
from the anonymous working of an invisible handls Instead, observed outcomes of
input use reflect the intersection of a parastatal’s decision to provide a household with
the input (and the crop) and the household’s decision to agree to the terms and condi-
tions of the input delivery.16

Following Poitier (1980), this entails the following model: the farmer wants to
work with the parastatal (i.e., to receive fertilizer) = y2‘, and the parastatal "accepts” the
farmer = y1‘ (here the ‘ reflects a latent or unobserved variable; we observe the
unstarred variable yi).
Parastatal decision: y1* - x81 + u1 [: 1 if yi* z 0, (for i=1,2)

. I
I

Household decision: yz* - x82 + “2 0 otherwise

In the model, we observe:

Z'Y'Y'
1 2 0 otherwise

2 could be a continuous (tobit) variable, e.g., if data exist on the kilograms of fertilizer

used. Poirier assumes a bivariate normal distribution in his analysis:

 

1SSee also the discussion in Binswanger and Rosenzweig, 1986, p.510.

16Moral hazard and adverse selection pose further analytical challenges to modelling
household input demand functions.

41

u 1 a
1 - iid BN 0, “
“2 au 1

Then P(Z=-1) = P(-u1 5 x81, 412 S sz)
= Fowl. .32, on).

where F is the bivariate standard normal density. From this a likelihood function can be
set up to obtain maximum likelihood parameter estimates.

Figure 2-5 does not explicitly show interdependencies among inputs. In addition
to technological complementarities among inputs, we may define a W M
W in environments where markets do not function well for some or all inputs
(including credit). For example, Kelly and Gaye (1986) argue that household heads can
only attract non-family workers and married sons (i.e., form coalitions) by providing
peanut seed, which traditionally has been received from the government on a credit
basis. 17 Hence a reduction in peanut seed credit may lead to lower availability of labor
to household heads. Similarly, household heads owning draft equipment may be in a
better position to attract non-family workers and / or retain married sons within their
households. The forming of coalitions among and within households is discussed in

detail in Section 2.3.

I Delgado and McIntire (1982), and Dione (1989) argue that farmers in the Sahel
will not adopt fertilizer and equipment unless they also have access to additional labor

due to the labor-shifting effects of new technology—e.g., increased weeding and harvest

 

17Kelly and Gaye also argue that this custom makes it difficult for household heads to
invest in improved cereals production technology (e.g., fertilizer) since whatever means
(cash) are available at planting must be devoted to the acquisition of peanut seed.

42
labor requirements. 18 To the extent that there are complementarities between labor and
equipment or fertilizer use, households with access to a larger labor force will be more
likely to use improved technologies, m m. Senegal’s fertilizer distribution
policy may confront reluctance from farmers if they cannot also hire more labor, for
example, by having increased peanut seed and food available to attract that labor at
planting. Since private sector credit is a severe constraint, it may be necessary to provide
farmers with short-term consumption credit so that they can provide additional laborers
with food. (The relationship between a household’s labor force and the use of fertilizer
is examined further in Chapter III.)

Marketed surplus elasticities for crops can be calculated directly from reduced
form equations, or as the (quantity-) weighted difference between demand and supply
elasticities of the good in question (see, e.g., Strauss, 1984, and Chapter V). In this
study, and consistent with the framework presented earlier, marketed surpluses are
treated as a valve for adjusting household-level stocks of food to cape with uncertainty,
rather than a planned, static quantity. Hence, net marketed surpluses depend in part on
beginning stocks (i.e., levels of output in period a).

Furthermore, given uncertainty and thin markets in southeastern Senegal, trans-
actions costs—manifest in a differential between buying and selling prices-are allowed for
when a good is bought or sold. With transaction costs, adjustment is no longer friction-
less, and marketed surpluses respond only to large values of predetermined variables.19

If Y denotes the actual change in stocks (i.e., net sales) and Y‘ the desired change, then

the relationship between these two variables is as shown in Figure 2-6 (Rosett, 1959). al

 

18Note that late rains during the peanut harvest are a problem for peanuts not yet
brought in from the fields; this means farmers cannot postpone the storage of peanuts
indefinitely.

19See Maddala, 1988, p.162, ff. for a discussion of this class of models.

43
(02) represents a desired decrease (increase) in stocks, and Y3 Y“ only for large (abso-
lute) values of the predetermined variables. The implication of recognizing transaction
costs is that economic adjustments have to be modelled statistically in two steps. First,
there is the yes-no decision of whether or not to adjust (and in which direction); this
involves a threshold, and a comparison of the benefits to and costs of crossing that
threshold. Second, a continuous decision is made regarding by how much to adjust,
conditional on having crossed the threshold. These concepts are discussed and applied
in more detail in Ch. V.

FIGURE 2-6: ADJUSTMENTS NITH FRICTION

/ > Y*
/°1 “2

(593m: Rosset, 1959)

 

 

As indicated earlier, the recursive nature of outcomes, and the associated uncer-
tainty, has implications for the ability of specific output price policies to effect produc-
tion increases. A higher official post-harvest food price announced before planting can
only affect total production if there is an intertemporal market (i.e., for credit), which
allows producing units to capitalize the higher expected future benefit into current-
period inputs, or if liquidin from other sources (e.g., investments in non-farm activities)
can be bid away to farm activities. For a production unit operating near subsistence and

having to decide whether to consume or to plant coarse grain seeds left in storage from

44
the previous year’s production, the prospect of a higher post-harvest price will offer little
immediate incentive.

The framework presented here therefore indicates that relative prices and fixed
factors are not the only variables governing economic behavior; production outcomes
and access to resources do influence consumption and marketing decision as well as the
acquisition of inputs such as labor through complementarities arising from market
uncertainty.20 Questions to be asked should include: Which households have which
resources and why? This contrasts with a static model where production and consump-
tion in essence take place at a single point in time. The conditional recursive satisficing
ESP model is also not the same as the 9; ant; maximization of a net present value
function over multiple periods, since outcomes for production and consumption varia-
bles are not assumed to be known a priori. More generally, the purpose of this expo-
sition is to describe and organize events and ideas (Thurow, 1984, p.23), explicitly taking
into account uncertainty and lags in agricultural production, consumption and invest-
ment behavior. This will lead to a richer understanding of economic behavior, for
example in the case of marketed surpluses, which would otherwise be summarized as a
point-estimate of an elasticity.

A further question to be investigated in more depth in this study is whether or
not there are cost savings when cash and food crops are produced within the same firm
rather than two separate (specialized) firms. This would further strengthen the argu-
ment that cash crop- and food crop-producing households are better off by forming a
household coalition and perhaps encourag'ng specialization within households. Figure

 

20This is similar to the constrained maximization problem developed by Chambers and
Lee (1986) and Lee and Chambers (1986), who use duality results to estimate an
expenditure-constrained profit function; while allowing for credit market imperfections,
these authors also assume perfect foresight, however. A similar (revenue-maximizing)
approach is found in Gordon, 1989.

45

FIGURE 2-7: COST SURFACE FOR TWO OUTPUTS

CW

 

 

(m: Baumol et al., 1982, p.84)

46
2-7 shows a hypothetical cost surface for a firm and three production possibilities fron-
tiers reflecting the trade-off between yl (the cash crop) and y2 (the food crop). Cost
concepts associated with multi-product firms are investigated in more detail in Ch. V
(see also Baumol et al., 1982). In this diagram the cost surface is "bent inward", indi-
cating it is cheaper to produce a linear combination of Y1 and Y2, instead of specializing
in the production of only one of the two products.

The estimated parameters of a cost function permit testing the hypothesis that
the existing farm technology—and not only transactions costs, uncertainty and input
complementarities—reduce the propensity of households to specialize into cash or food
crop production. Estimates of product-specific scale economies provide information as
to where households are currently operating on their cost curves, and how the average
household’s costs of production will change if crop substitution takes place as planned
under the NAP. For example, if the per unit cost of producing food increases, there will
be important implications for net buying households.

In summary, the analysis in Section 2.1. was based on a strict trade-off between
food and cash crop production. The discussion here raises the possibility that there are
nevertheless certain complementarities between food and cash crops, to the extent that
costs (resources) are saveduor inputs used more optimally—when both are produced on
the same farm. In a broader sense, the two types of crops may also complement one
another in that cash crops, with their assured market outlets, allow farmers to pursue
investments in resources such as draft technology. Hence dynamic elements come into

play which may not be picked up in a static analysis.

47
23. RURAL ORGANIZATION: FARM HOUSEHOLDS AS COALITIONS

[I]f... the differences between LDCs and the more developed countries

lies largely In matters of economic organization, then the first item on the

research agenda should be a better understanding of the W

of LDCs. What IS needed 18 a theory of rural organization, as well as a

theory of industrial organization focusing on the special characteristics of

the LDCs (emphasis in original).

LE. Stiglitz, 1989, p.202.

Uncertainty and transaction costs affect the ability of markets to function
effectively. Alternatively, they may also lead to the formation of institutions such as
coalitions, manifest in the form of integrated or extended households. That possibility
and its implications for economic analysis is examined in this section.

Agricultural households-as opposed to individual workers or villages-are
appropriate units of observation for economic analyses since they are generally the level
at which production, consumption, land allocation and resource investment decisions are
made.21 Households are not composed of individuals with identical status, rights and
goals, however, and it is important to understand how the various actors cooperate (or
fail to do so) in a household coalition.

A general and workable definition of an agricultural household in Senegal from
the production side is provided by Benoit-Catfin and Faye (BCF, 1982, p. 83)22:

L’exploitation agricole familiale est l’unité de production constituee par

l’ensemble des membres d’un groupement familial qui partagent la méme

cuisine et dont l’a‘iné W en y affectant une partie de sa

production en contrepartie du travail que lui allouent les autres membres
du groupement. Le reste du temps de travail est utilise librement pour

 

218cc Casley and Lury, 1982, for issues involved in defining farm households in
traditional agrarian societies.

2213mm: The agricultural household family is the production unit made up of all
members of a family group which shares the same kitchen and for which the eldest member
W by providing part of his output in exchange for the labor of other
group members. The remaining labor is used freely to cultivate land belonging to the group
and managed by the elder, and the output of which is individually appropriated.

48

cultiver sur les parcelles appartenant an groupement ebgerées par l’ainé

et dont la production est appropriée mdrvrduellement.
Implicit in this definition is a creditor-debtor or patron-client dependency between the
household head—or m ndjgu], the ”owner of the kitchen"-and his W (see also
Monnier, 1974; Diop, 1985).24 The word ERIE derives from m: to satisfy hunger and
ga: to cause to work.25

A further useful concept for understanding dependence among relatives is that of
household integration or extension. Binswanger and McIntire (1987) define two forms
of household extension (p.81-2):

(1) a vertically extended household, composed of nuclear units of succee-

ding generations, and (2) a horizontally extended household, composed of

nuclear units of siblings. ...[A] household can be both vertically and

horizontally integrated.
These relationships are shown in Figure 2-8, which also defines household-related terms
used in the present study. In the vertically extended household the head is called a _c_h_ef
W191], while in the horizontally extended case he is a Wu (CC).
W consist of two or more horizontally integrated gums, wherein the CC
(usually the oldest brother) allocates land belonging to the W among the
independent W. While the latter make their own production decisions, they

are therefore constrained in terms of their access to land through their membership in

 

23FSP experience shows it is not necessarily the oldest person who decides. Monnier
(1974, p.37) quantifies labor exchange flows in one household.

24BCF (p.35) use the term samba to include all dependent household members, i.e.
[1mm (those that come during the ggwet, or rainy season), m (defined is this
study as non-family workers who reside permanently in the household), and family members
including wives. Was were actively recruited during the colonial era (Ba, 1986, p.214).
The My; make up another category of (young male) migrants, arriving in the Peanut Basin
during the peanut harvest season (Ba, 1986, p.210, 214).

25BCI-‘, m. Literally, the word means to feed someone in exchange for that person’s
labor.

49

FIGURE 2-8: VERTICAL AND HORIZONTAL EXTENSION OF HOUSEHOLDS

ll"

 

 

 

E-Exploitation

C-Concession
Gij-jth married son
of the ith generation

5

 

 

—E
‘ G21

 

 

 

 

 

 

 

 

 

GU

E

 

G22

 

__.E »—E

623

l

 

 

 

 

 

 

 

(1) Vertical
Extension
[1 decision-making
unit]

(ii) Horizontal Extension
[4 interdependent
decision-making units]

the W. A possible benefit to being a member of a concession may be that it

facilitates the borrowing of agricultural equipment.

To fulfill the requirement of feeding his dependents (both relatives and non-

relatives), the household head (ghgf mm cultivates a collective cereals field

(BCF, 1982; Diop, 1985; Goetz and Diagana, 1987), to which each dependent (flung)

supplies a predetermined amount of labor.26 This amount can differ by dependency

status, household, region and ethnicity. Proceeds from the collective field can also be

used to purchase clothing for dependent workers and, when all other needs have been

met, for the household head-who generally also owns personal fields to meet his cash

requirements. To some extent a measure of the degree of solidarity of the household

(Diop, p.176), the collective field is generally closest to the household (Figure 2-9), and

at least in the Peanut Basin is continuously cropped to early-maturing millet which

 

26See also Stomal-Weigel, 1988, p.23. He reports that some villages in the Peanut Basin
have millet fields which are collective at the level of a village quarter (m).

50

FIGURE 2-9: DEPENDENCY STATUS AND LAND USE

 

 

 

 

 

 

 

Ken: 191m): a
Case Champ de Case Champs de Cultures Champs de Brousse
House Household Field Crop Fields Proper Bush Fields
-hold
//
”/1
Household (collect.) FPM FPMP FPM FPS
Crops: Continuous or or or or FP
Cereals FMP ZPSP FMP FSP
\\
Household head; dependent HH heads

 

 

.4
Hives and surghgs
.<

a. Proprement dits.
Crop rotation legend: F-Fallow, P-Peanuts, M-Millet, S-Sorghum, Z-Maize.

(Sggrgg: Monnier, 1974, p.25.)

51
benefits from animal manure and/or ”night soil".27 As discussed in Ch. IV, the distinc-
tion between ”home maize" (mi; g; m) and "field maize" (mm d; ghamp) is
important. The potential for expanded production of the former is limited while-
depending on the region-lack of soil fertility and attacks of wild animals constrain the
expansion of the latter.

In the cash crop-food crop debate it is often argued that farmers face a trade-off
in terms of producing both types of crops. In this regard it is interesting to note that
BCF (pp.69-70) argue that the introduction of expanded28 peanut production led to an
extension of the work day and work week, eliminated certain labor slack periods, and
attracted migrant workers. They also contend that peanuts displaced cereals formerly
used for sale rather than consumption, and therefore did not disturb the household’s
food balance.

Stomal-Weigel (1988, p.29) states that the household head in general controls the
production of the food crop and has the right to manage it, but does not own it. Accor-
dingly, this constitutes one factor preventing food crops from becoming more of a "cash

crop”29:

 

27See Monnier, 1974. Informal survey evidence suggests this constellation of fields is
also prevalent in the southern research areas.

28At the turn of the last century peanuts were cropped only in small quantities that
were destined solely for on-farm consumption. This raises the question of which mechanisms
were put into place so that peanuts could become more of a commercial crop.

29This argument pertains to Wolof and Serer ethnic groups in the Peanut Basin. It is
not known to what extent the results can be transferred to other groups, such as the Peulh.
flnnslatigg: Being a collective good, [the household head] does not dispose of the millet
(food crop), being a collective good, at will and in theory cannot sell it without the group’s
consent; if he does frequently sell it in practice, it is because he can assure the food supply
of the collective with other means. Therefore, given conflicting social needs within the
household, and each individual seeking to meet his own endogenous objectives through sales
of the commodity, millet must lose its collective character and become an individual
commodity;

52

Le mil étant un bien collectif, [1e chef d’exploitation] n’en dispose pas
librement et, théoriquement, il ne peut pas le vendre sans 1e consente-
ment du groupe; s’il le fait souvent en pratique, c’est parce qu’il est en
mesure d’assurer autrement les besoins en vivres de la collectivité. Or,
pour devenir une production commerciale au sens strict du terme et étant
donné les contraintes sociales en presence, chaque individu ayant ses
objectifs endogenes a satisfaire grace a une participation a la production
marchande, le mil doit perdre son caractere collectif et devenir un bien

individuel;

Most dependent household heads (married sons) grow food crops so that their
wives can supplement the food provided by the household head when it is their turn to
prepare the meal of the household (BCF, p.76). Married women have their own
personal fields, the proceeds of which are used to purchase condiments for cooking and
household items, and to satisfy personal needs.

Stomal-Weigel (11214.) maintains that peanut sales are not well-suited for meeting .
daily cash needs of females, since the peanut marketing season is of limited duration and
peanuts are difficult to store (his observations are based on experiences in the Thies and
Diourbel regions). Therefore (p.29),3o

[c]e sont surtout les femmes, ayant une obligation de founir

quotidiennement les condiments nécessaires pour la preparation des

repas, qui apprécient la régularité des revenus que leur assure la petite

parcelle cultivée en mil.

It is not obvious how to measure the welfare of females or wives. In rural
Senegal, marriages are interpreted as ”alliances" between different families (Diop, p.206)

31

and should not be confused with Western notions of matrimony. The argument is

sometimes made that a wife approves of the husband marrying further wives since this

 

3mm: It is mainly the women, with the daily obligation of providing condiments
necessary for preparing meals, who appreciate the regularity of income provided by their
small cereals fields.

31h is known that marriages are sometimes arranged over space to spread climatic risks
(eg. Binswanger and McIntire, 1987, p.84). In the FSP sample at least one matrimonial
alliance was made across the Gambian border, ostensibly to facilitate the introduction of
cheaper Gambian products into Senegal. See also Lambert, 1989.

53
reduces the average burden of household chores.32 First wives have a somewhat higher
social status than subsequent wives, but this does not necessarily extend to economic and
other spheres. Wives also do not wish to share their husbands with too many additional
wives, and may encourage their sons to become autonomous soon after their own
marriage; this labor withdrawal from the household head reduces his ability to take on a
new wife. For the same reason, the household head may not want to provide his son(s)
with large amounts of peanut seed; BCF (p.73) also suggest that acquisition of equip-
ment by sons will hasten their emancipation, i.e., departure from the household. A high
dowry-viewed as an economic exchange for the procreative ability of the female-can
make the decision of whether the household head or his son obtains the next wife a
source of conflict in the household (Diop, p.195).33

BCF (p.79) suggest that women are better off if they belong to large, fully
equipped households and have sons who work their fields.34 This can mean that the
mother need not work agriculturally, while still benefitting from the output of her alloted
field to meet her own needs and communal obligations. She may of course also receive
transfers (gifts) from her sons’ fields and future in-laws.

The difficulty of mobilizing rural deposits in Africa is well documented (von
Pirschke et al., 1983; Binswanger and McIntire, .1987). Diop (p.170) describes the
situation where young males aspiring to become household heads deposit part of their
surpluses for safe-guarding with the household head (in which case the son receives an

interest payment) or, when confidence is lacking, with a maternal uncle or the village

 

32Polygamy, with no limit on the number of wives so long as each could be supported
adequately and equally, existed prior to the introduction of Islam into Wolof society.

33The size of the dowry also discourages divorce, which is one manifestation of the
breakdown of a household coalition.

34Obviously, there are also tangible economic benefits to having female off-spring.

54
chief. Similarly, dependent unmarried males can grow food which they deposit with their
(older) married brothers or household head, and which is credited towards a future
dowry payment made by the recipient on behalf of the depositor (BCF, p.48).

In addition to household production units one can define residential, consump-
tion and accumulation units (translated from Gastellu, 1980, p.4):

1. a residential unit—the group of persons sharing the same living space,

separated from others by a visible frontier;

2. a production unit—the group of persons contributing to the creation and

provision of a product;

3. a consumption unit-the group of persons participating in the destruc-

tion of part of the product with the goal of regaining their ability to work;

4. an accumulation unit—the group of persons who collectivize the surplus

left over after each individual’s consumption.

While, in general and for practical purposes, the first three types of units are
often the same, it has been observed among certain ethnic groups that independent
households of a concession form a collective consumption unit during the rainy season.
These ethnic groups include the Serer of Niakhar in Senegal’s Peanut Basin (Lombard,
1987, p.474), and the Serer Safen in the Ivory Coast (Gastellu, 1980, p.7). It is not ob-
vious how this phenomenon would affect agricultural household analyses, where utility-
maximizing production and consumption decisions are assumed to be made simul-
taneously by a single, well-defined and homeostatic unit.

Of more immediate interest is the phenomenon of an accumulation unit (AU)
among the Serer of Mbayar in Senegal studied by Gastellu (1985, p.417). His analysis
may have implications for the disinvestrnent behavior of households in case of droughts.
AUs are composed of ”uterine relatives" such as a household head and his brothers, or a

mother and her offspring, who accumulate wealth in the form of livestock, agricultural

55
equipment and jewelry. Three purposes of this accumulation are to provide (1) produc-
tive capital; (2) a security stock; and (3) a means of ceremonial exchange between AUs
(Gastellu, p.419).

More importantly, AUs are not neatly confined or localized within a household,
but are spread across difierent households. This leads to an “institutionalized contradic-
tion,” since each individual has obligations relative to his or her production-consumption
unit and a separate AU, which in turn prevents pronounced individual accumulation
(Gastellu, p.417r8). In the two societies studied by Gastellu, the contribution of indivi-
duals to the AU remained recognizable and separable. With geographic dispersion, AUs
may entail risk-spreading benefits, and affect the ability of different households to cope
with climatic uncertainty.

In case of food production shortfalls, the household head can purchase food from
his dependents using proceeds from his cash crop fields, or he can buy cereals from his
wife with reimbursement at the next harvest, with the amount of interest charged by the
wife depending on the degree of competition between wives (Diop, p.166). Various
authors report that output from the collective field is allocated first to consumption in
the subsequent rainy season, and the remainder (if any) is then consumed during the dry
season. Diop (mid) writes that when the household head migrates in search of food, the
wife may have to sell ofl' personal belongings to feed remaining household members in
the interim. The first wife is responsible for the household’s upkeep when the head is
absent, and also represents the other wives in front of the household head (Diop, p.189).

Diop (p.214) presents data showing that of 921 cases of litigation before courts in
the Thies, Diourbel and Dagana regions during 1965-73, 36.4% involved household
conflicts (including the effects of divorce) while 37.5% concerned divorces, or a break-
down of the coalition. Among reasons given for divorces, the inability of the household

head to provide food was prominent (p.223). According to the same author (p.180) this

56
is also a common cause for dependent household heads to form their own independent
household.

The labor benefit which the household head receives is the counterpart of a
transaction requiring him to supply dependents with food, lodging. land and (variable)
inputs. As Binswanger and McIntire (1986, p.83) point out, "labor incentives are clearly
improved if all nuclear units, or even all members, are also allowed to have plots of
their own.” The inputs supplied to dependents may vary with the status of the dependent.
BCF state (p.77):35

[1e chef d’exploitation] doit assurer les investissements nécessaires aux

activités agricoles du foyer et pour ses propres parcelles les dépenses

annuelles de culture: engrais, semences, etc.

Diop lists the responsibilities of the household head as (p. 158):36

Détenant seul le pouvoir économique, [le chef d’exploitation] a, en retour,

l’entiere responsabilité de l’entretien de la famille, des l’instant que ses

membres s’acquittent correctement des taches - surtout agricoles - qui

leur reviennent. Il doit fournir les instruments de travail, trouver des

semences, s’il en manque, assurer la nourriture, meme en période de

soudure ou de disette.

The requirement of providing inputs and food to dependents may partially
explain why one particularly successful and large-scale farmer had difficulty envisioning a

doubling of his operation (suitable land was not a constraint). Household heads not able

 

3SILanslatigm: The household head must assure the necessary investments for the
agricultural activities of the household and the annual expenses for his own fields: fertilizer,

seeds, etc.

361nm: The sole holder of economic power, [the household head] in return has
the entire responsibility for the upkeep of the family as soon as the other members have
properly carried out their - mainly agricultural - tasks. He has to furnish tools, find seeds
if they are lacking, and assure the supply of food even in periods of food shortage.

57
to meet their obligations face the threat of other members exiting from the coalition.
For example, Diop states (p.159):37

([m]éme dans le passé,) des chefs de ménages quittaient la concession
pour fonder la leur, non seulement quand la famille devenait trop grande
cu que les terres manquaient, mais aussi lors-qu’ils mettaient en cause
l’autorité du chef de concession qui abusait de son pouvoir on W

W. I] y avait aussi une alternative pour les jeunes gens
qui pouvaient rejoindre la maison de l’oncle maternel (emphasis added).

Further, (p. 176)38

[a]ujourd’hui, l’autorité du chef de famille depend avant tout de sa
puissance economique mise au service du groupe et de ses membres. Les
Wolofs disent de lui: ”S’il ne peut pas satisfaire leurs besoins, ils ne le
respectent plus."

Discussing recent changes in intra-household relationships, Diop also argues that
traditional m (dependents) are becoming more independent because of increased
participation in the monetary economy, due primarily to the availability of peanut
production (p. 171):39

[l]e pere de famille ne peut s’y opposer, meme s’il le voulait, en lui

refusant la disposition d’un champ personnel par exem'ple. il risquerait

de provoquer le depart du gargon - qui irait s’établir ai'lleurs, chcz son

oncle maternal - faisant perdre ainsi a la famille une force de travail

appreciable. C’est une reaction qu’on pouvait noter, autrefois, en cas de
conflits graves dans la concession.

 

 

37W: (even in the past) dependent household heads would leave the household
to create their own not only when the family became too large or land became a constraint,
but also when they challenged the household heads authority if he abused his powers or
.One alternative for the young individuals was to join

the household of their mother’s brother (emphasis added).

38113115139521: Today, the household head’s authority depends primarily on the
economic strength he conveys to other household members. The Wolof say of him: "If he

cannot satisfy their needs, they no longer respect him."

39mm: The household head cannot oppose this behavior [of economic inde-
pendence], even if he wanted to, for example, by refusing to provide him [the dependent]
with a personal field. he risks provoking the departure of his son - who would settle
elsewhere, with his mother’s brother - thereby causing the family to lose a worker. This is
a reaction noted earlier in the case of serious conflicts in the household.

58

The central role of peanuts in the household coalition is further stressed by
Stomal-Weigel (p.28):40

GrAce a la production de l’arachide, chaque individu peut' réaliser ses

propres objectifs indépendamment des objectifs de l’unité de production

en tant que telle. Le chef de l’unité de production garde toujours le droit

d’affectation de la terre parmi les invididus (s3) et selon les cultures; ce

faisant, i1 doit cependant tenir compte des objectifs des différents groupes

restreints et individus de son unite de production, au risque de perturber

l’équilibre du systeme.

Finally, BCF (p.10) report an interesting consequence of the labor arrangements
within households for the adoption of a new cultural practice. Plowing under plant
residues of cereals at the end of the rainy season appears to appreciably afiect yields in
the subsequent season. However, household heads were found not to adopt this practice
because once the short-cycle collective millet crop had been harvested, they no longer
benefited from the labor services of dependents.

As a summary of the preceding discussion, Table 2-3 provides an overview of the
objectives, rights and obligations of various members in Wolof households or coalitions.
The principal objective of the householdusurvival-requires both a reproductive capacity
and the ability to ensure adequate food availability. It is probably safe to argue that the
table is more or less valid for other agriculturally sedentary ethnic groups of Senegal.
Certainly among the Peulh of Senegal, the household head or 19m gang is the central
decision-maker (Ba, p.139), and vertical or horizontal integration are institutions also

encountered in the southern research areas.41

 

4012mm: Due to peanut production, each individual can achieve his or her
objective independently of the objectives of the production unit (PU). The head of the PU
retains the right of allocating land among individuals and crops; in so doing he must,
however, take into account the objectives of the different groups and individuals of his PU
so as to not perturb the equilibrium of the system.

41One southern village, for example, consisted of 11 households with only two family
names: "MBallo' and "Balde". All the MBallos (6 households) had formed separate
households while the Baldes (5 households) had organized themselves within a concession.

.59

.ooum.p zooms—u omega cos» cocoa muaac_ a "one:
.aeocn cocoons; emu ecu scam oozep.>o pocc.uo>comnc co comma a. .~.~
..ea.n .~om_. exec nee e.eueu-u.eeem no sense »_ee.eee one .m.~-~.~ .»__e.ueea n. ._ "tmummm

 

acumeou a.o:un
tan: o\3 u.oocu
o>.oooc acccou u.ooeu .~.~

 

aucopcoaou cu manna. __u noc.>oca nausgc~ .o.~

 

 

 

 

 

 

 

u—oms_z e.gao oo.>coa .ueeo
o» mc_cc.mom oo¢>oEa mu_oz gnomes emu :. mo.co>.
A ttttttt u:oEn.:uo a.uoo;:; ou:x:becon ....... v accen_:ao on: Lo :30 «on on u:u2n.:ao era L.o;u :zo u:o§n.:uu .m.~
.uoogzg .cozuos «man A we use v v.0.» ensues oo—o_e opus co
Logan» no» xcoz a.uoon uposomao: co exec: Lo; aua.»nu Lone. pauses nope—uxnnoco are a.: no exact gonna .v.~
«new: »_.26» com xoeo.o_meaatu—oo can a—ooEoo «roan
noou mean «seen A.nno eon. o>ona wanna. penny—zo.cmu new
gene anew: smog —e:oaaoa Lou Aa—oocoo coco nou'o «use: Lou «use: case peeoacoa no» .mpoocou
o: coauoo .aaacooa. coco sane «scan a: coco sane macaw cauuou .auncooa. coco sane axocn necu .m.~
epone.s a. menopo
use. co ». au.conc. u. veep :30 one. accu u o>oc upogoaao: copnmoocou
on agape a: use Amoco o o» ucn.c on» no: Ou usm.e o: o» u:u_e were nos» we use. nuances use; .~.~
poms ocoaoca cu mucoe_u
can» cos a. a. taco no.pnn=m
cos; too» so.) Amo>.3 ocean «pao.u:. sconou.xe «.5 no
ou.: nou_>0ea canoes ”cannon nasty ”census can can poem mop—>ocn
A ttttttttt one; upogomso; onu an on» ttttttttt v so; auu.mma pans «cannon; axons” poem. acococu a «:30 need .~.~
mzc_p<w_4mo
az< m»:u_¢ .~
3263-3 $0 2833—.» 3%: ~28... 6.: Costa:
.cac usage—om can mo>_3 mo Eunsac one o~.s_xoz
A~oaaoam one. «use: —o:0ncoa com .o:_ sane :.ouao
scene. .me eec naee2.eeee a .e.ec=m-c_em neon
ouzoan Lou "gonna _uu.nou mono: opus: a .o.uu=mtu—om toad ”copmmuocco co\a m>_huuaao
"soaps pea—neg neon:— zmau pocomcoa soc "upononsoz no upogouao; Co u_xozouu
oaouc~ zaou coco poo» meow «sou:— saou po>.>c:n ocaaa< pu>_>c:» ocaaa< >¢<x_¢m .—
xoa ope: uo_ccu2:: one: u—ogoaao: o—asou opgsuu one: upogouao: co.naooccu
o>.uo< mesa» acoucoauo unoccunoa o>.uo< acne» to.cco: unoccoaouc_ we use:

 

uo~hua¢auxaaazum=o= mean: “2» a— mac—b<w—4¢c 58¢ mhzw—s .mw>~huw6¢c "mun man<h

60

These observations in principle suggest that farm households in Senegal can be
viewed analytically as coalitions of individuals with divergent interests: the household
head, responsible for the long-term survival of the nuclear family is preoccupied with
growing sufficient cereals (and presumably some cash crops to satisfy household cash
needs), while wives, young sons and migrant workers are concerned with generating cash
income. More specifically, the household head primarily seeks to maximize food
production to ensure the survival of his household; the wife seeks cash income for
personal and household needs as well as a balanced number of co-spouses; the married
son seeks to accumulate sufficient capital (cash) to eventually form his own household;
and the non-family worker, similarly, seeks to maximize the amount of cash received at
harvest. Given the importance of cash income, it is interesting to speculate whether the
cash-food crop dichotomy would ever have arisen in Senegal if food markets had histori-
cally been as reliable as cash crop markets continue to be to date.42 A further general
objective of the household is to support young children as well as elderly and infirm
members of the household. This implicitly provides a form of intergenerational
insurance.

The above problem can be stated more formally as follows.43 Let {N} a
{H,F,M,W} = {1,2,...,n} denote a set of individuals capable of forming a household
coalition. They may include a household head (H, or the manager of the coalition);
females (F) who are old enough to be married; the household head’s sons (M) who may

or may not be married; and non-family laborers (W), where each group pursues a

 

42Cash crops can always be sold to parastatals, although farmers may not receive
officially announced prices due to opportunistic behavior at the scales. It is interesting to
recall that peanuts were sold solely as a food crop at the turn of the century.

43The discussion is based on Staatz, 1983, who examines how costs should be shared
among members of a cooperative so as to ensure its stability; his conclusions also hold in
the present context.

61

different set of objectives. Further, let C(qm) measure the total cost facing members of
subset M if they were to provide the amount qm of a service to themselves, and
C(qh+ f+m+w) denote the total cost at which H, F, M and W could together obtain
(qh+ qf+qm+qw) of the service.44 Then there are economies in jointly providing ser-
vices to the members of the coalition, or subadditivity of the cost function is assured if,
foranysubsetsH,F,MandWCNandHnFnMnW = (a,

C(qh+‘+m+W) s on") + C(q‘) + as”) + cm")

A member of group j can obtain qj from the coalition or attempt to obtain that
quantity elsewhere; for example, the married son may seek off-farm employment if he
can thereby obtain (at least) qj at a lower (or equal) cost. Obviously, the analysis
assumes perfect certainty and foresight with respect to production outcomes. If the rains
fail, so may the coalition to the extent that actual benefits received differ from those
initially contracted for.45

Further following Staatz (pp.1085-6), a characteristic function V(qj) can be
calculated, with knowledge of the cost j would incur if it were to obtain the service
elsewhere, for each subset of members to reflect the minimum cost of providing qj.
Stability of the coalition is assured if A(qm) < V(qm) for all M c N. Here A(qm) is
the share of total (coalition) cost borne by subset M. We can also calculate the (net)

benefit obtained by member M from participating in the coalition as (equation 5 in

 

Staatz),
C(an)
m m
bm a V(q ) ' qn q

 

44F or example, the ”service" may be kilograms of peanuts harvested for sale to
SONACOS.

45 During 1987, non-family workers departed one research triad because of a
(temporary) drought during the production season.

62
We may then ask which variables affect A(qm) or bm in the context of Senegal
(Table 2-4), and how they are in turn affected by agricultural policy. Obviously, any
policy which increases the cost of peanut seeds, such as an elimination of the peanut
seed credit program, can cause the coalition to collapse and lead to a reduced household

TABLE 2-4: ALLOCATION OF COSTS AND BENEFITS IN THE COALITION

Contributions (-) to and

 

 

Benefits (+) from the Household Females [Married] Non-Family
Coalition Head: H F Sons: M Labor: N
Land -(l 1 +1 ) l l 1
Draft Equipment & Animals -(d;:d$+d:) d; d: d:
Peanut Seeds -(pf+p +pw) pf p pw
Food Consumption -(ff+f$+fw) ff f: fw
Labor for HH Cereals Field cf+cm+cw -cf -cm -cw

 

labor force. The table shows who gets what from whom in the coalition. For example, If
shows the amount of land females receive for growing their crops.

The idea that a coalition of individuals can produce a set of g goods (or services)
less expensively than can g individuals on their own has of course
existed for some time, and is central to the theory of cooperation (see also Olson, 1971;
and Axelrod, 1984). The concept will be used again in discussing economies of scope in
food and cash crop production.
2.4. SUMMARY AND CONCLUSION

Within the context of informing Senegal’s New Agricultural Policy in general and
the cash-food crop relationship on farms in particular, what do the concepts just
presented suggest in terms of a useful line of inquiry and data analysis? First, while the
concept of the household as a singular decision-making unit remains a valuable

abstraction (and will be used extensively in this study), the notion of a household as a

63
coalition of individuals with divergent interests suggests it is important to identify what
difi'erent types of household members get in terms of inputs, what they do in terms of
production (i.e., what economic forces motivate them), and what difference the organiza-
tion of a household makes for food production performance and the design of policies
intended to raise food production levels. Also in terms of the NAP in general and the
APS in particular, what are farmers’ investment priorities and their perceptions of
mineral fertilizer use? A related question is whether variables such as access or
entitlements to cash crop parastatals help in "explaining" equipment ownership.

Second, if there is a market-based interdependency among inputs, a complemen-
tary rather than strictly competitive relationship may arise between cash and food crops.
In particular, the peanut seed obtained on credit may allow the household to attract
additional labor for work on the collective cereals field of the household. These input-
related questions are addressed in Chapter 111.

What difference does access to and the use of agricultural equipment make?
From the farmer’s viewpoint, which non-price factors constrain crop (especially maize)
production, and which non-crop income generating strategies do farmers pursue? In
addition, what can be learned from multi-product household cost function estimates
about the consequences of inducing farmers to switch from cash to food crop produc-
tion? These questions are examined in Ch. IV.

Given environmental constraints and tastes and opinions of farmers, how are the
new policies likely to be received in rural areas and what is the likely potential for their
success? How do the strategies identified in Chapter IV affect the food consumption
security of households? Finally, can specific consequences of uncertainty and transac-
tions costs for the marketing behavior of farmers be identified? Chapter V addresses

this set of post-production issues.

CHAPTER III
HOUSEHOLD ORGANIZATION AND THE USE OF INPUTS

This chapter describes the organization of households and their 1986 resource
inventories and use of inputs to provide baseline information for Senegal’s New Agricul-
tural Policy. Household heads’ investment priorities at planting in 1987, and their
anticipated responses to the new input distribution policy are listed, with particular
attention given to concerns surrounding the use of mineral fertilizer on coarse grains.
Issues involved in the privatization of input marketing are identified using survey results.
Two econometric equations, designed to test for interdependencies among inputs (as dis-
cussed in Ch. 11), are presented. Policy implications from this analysis for the privati-
zation of agricultural input distribution functions in the context of Senegal’s food self-
sufficiency objective are briefly identified, and then will be expanded upon in Ch. VI
(Section 6.1.1).
3.1. RESOURCE INVENTORIES

The structural organization of households and the amount of different types of
labor available to household heads are described in this section, using a north-south
sample stratification. Information on draft equipment ownership and use as well as
chemical input and seed use in 1986 is presented. The current role of the private vs. the
public sectors in assuring input supplies and enabling investments in equipment is also
examined.

3.1.1. Structural Organization of Households and Labor Supply

The basic unit used to enumerate nuclear agricultural households is an @1933:
£an (defined from the production side). Emlojtatjons are generally independent
decision making units which may or may not be organized within-oar affiliated with—a
concession. As discussed in Ch. I], the key factor uniting the otherwise independent

64

65
95mm of a W is the allocation of land among the m (usually)
by the father or oldest brother who heads the Mm. It is also hypothesized that
there are other bonds or linkages between such exploitatigns, for example in terms of
food security strategies and access to resources such as equipment. W thus
consist of two or more horizontally integrated mm (cf. Ch. 11). Table 3-1 shows

the distribution of W and W in the census sample.1 For example, 73

TABLE 3-1: NUMBER OF W
PER ONC SSION

 

 

 

 

 

Number Number of

of Concessions
Exploi-
tations North South Total
One 166 160 326
Two 25 25 50
Three 5 12 17
Four 1 4 5
Five 1 1
Total 197 202 399

 

Sgurgg: Expanded Census (1986)

W consisted of two or more Mm while one concession had five
W. There is no significant difference in this distribution by ethnicity. In the
sample of 215 households used for subsequent in-depth surveys, 22% of the households
were horizontally integrated (HI).

Household heads who retain their married sons within their households as
dependents are heads of so-called vertically integrated households (VI). Of the house-
holds in the final sample, 19% were vertically integrated, while 9% were integrated in

 

1See Table 1-2 for a brief description of the census instrument.

66
both the horizontal and vertical senses (HI‘VI).2 Thus, one-half of the household heads
were involved in some form of a coalition with their off-spring and/ or siblings.

Another aspect of the coalition involves the use of non-family labor. Northern
households were more likely to employ non-family labor (Table 3-2), possibly reflecting
higher endowments of equipment, cash crop seeds (peanuts) and food. Non-family
workers are comprised of surfing and W the former have a stronger bond with
the household head, residing in the household throughout the year. The latter are
present during the rainy season (or 112119!)- Relative to their southern counterparts, nor-
thern households on average also had one more adult equivalent producer (AEP) at
their disposal (Table 3-3). As discussed in Goetz and Holtzrnan, [GH henceforth] (April
1988, p. 14-5), households in the southern areas on average gained .48 additional workers
(concentrated in 36% of the households), while northern households gained 29 workers
in 1987 relative to 1986. While the reasons for these gains are unknown, they may be
related to higher expected rainfall following a good year (1986) and discouragement by

TABLE 3-2: HOUSEHOLD LABOR FORCE DISTRIBUTION

 

 

Labor Percent of Households Total

Category North South Nbr. %
Female Labor 83 96 179 90
Young Males 62 56 115 58
Older Males 35 26 59 30
Married Sons 28 26 53 27
Non-Family Lab. 47 11 51 26

 

Sggrgg: ISRA/MSU FSP Surveys, 1986/7

Note: For a total of 198 households; based on FSP
field surveys, 1986/7. See also definitions
following Table 3-3. Standard errors are less
than or equal to .06 (i.e., 6%) in all cells.

 

the distribution is similar for the northern and southern areas.

67

TABLE 3-3: AVERAGE LABOR FORCE STRUCTUREa
(Standard Errors)

 

 

Category North South Total
Household Head .91 .94 .93
(.02) (.02) (.02)
Females 1.72 1.89 1.82
(. 16) (.12) (.09)
Young Males 78 .80 .79
( 09) ( 10) ( 07)

Married Sons .40 .40 .40
( 08) ( 07) ( 06)

Older Males A .30 .37
( 09) ( 05) ( 05)

Non-Family 1.08 .14 52
Labor ( .16) (.04) (.08)
Mean 5.37 4.46 4.83
St. Error .36 .23 .20
Minimum 1.00 1.00 1.00
Maximum 15.10 13.90 15.10
Valid N 81 117 198

 

Sggrgg: ISRA/MSU FSP Surveys (1986).

a. Data are adult equivalent producers (AEPs),
using weights: males 15-60 yrs - 1.0;
females 15-60 — 0.8; boys 5-14-0.5; girls
5-14-0.25; individuals older than 60 - 0.5
adult-equivalents; only active individuals
are included in calculations. ”Young males"
are 0-19 years, “older males'I are 20 and
older. The (agricultural labor) weight for
females is less than that of males to reflect
their preoccupation with non-agricultural
domestic work. This variable is intended to
measure the agricultural labor available in
households.

68
job seekers with alternative income opportunities due to cutbacks in public sector
employment.

Figure 3-1 illustrates the notion of a household as a coalition of individuals with
different (cash and / or food crop) interests. Married sons allocated their land to cash
crops more frequently than if they had headed their own household, while non-family
workers exclusively grew cash crops.3

While Figure 3-1 suggests females play a major role in growing cash crops
(especially peanuts), Table 3-4 shows the average quantity seeded by that group in the
north is only one-tenth of that seeded by household heads, and about one-fifth that
seeded by dependent married males. Similar relationships hold in the southern research

area, where females on average also tend to produce more cotton than in the north. It

 

FIGURE 3-1: CROPS GROWN BY HOUSEHOLD MEMBERS
(PERCENT OF FIELDS CULTIVATED BY EACH MEMBER)

 

10° V LEGEND

BO 0 § .\\V Cotton

80 J, >§ g Peonute
a 7D 0 a W 2::ne Grolne
E. 50 o
g 50 'i’ ¢
E 40 .h /
g 30 " A

20 .. 7

g

 

 

 

 

 

 

1 Other' Mole Non-FfWorls.
Hrd. Son Femolee
HOUSEHOLD MEMBER

 

 

Source: FSP Surveys (19I6/7)

 

 

3Land area data were not collected due to measurement difficulties.

69
TABLE 3-4: DEPENDENCY STATUS AND KILOGRAHS OF SEED USED; BY REGIONa

 

 

 

 

Dependents

Study Area
and Type Household Females Married Other Male Non-Fan.
of Seed Heads (Hives) Sons Relatives Labor

---------- Kilograms of Seeds Planted-----------
Neath:
Coarse grain 31 0 2 0 0
Peanuts 293 29 133 51 105
Cotton 5 l 3 1 2
Valid N 87 72 23 62 39
Coarse Grain 4o 1 4 l 1
Rice b 11 9 0 0 0
Peanuts 88 14 27 15 24
Cotton 15 3 17 5 7
Valid N 124 119 32 83 13

 

figures: ISRA/MSU FSP Surveys, 1986/7
a. Data are per adult equivalent producer (see Table 3-3).
b. Unshelled peanuts.

is of interest to note that although households heads are responsible for growing coarse
grains, they on average also seed the largest quantity of peanuts. Married sons grow the
largest fields of cotton in the southern research areas. The question of seed inputs is
examined again in Section 3.1.4.

3.1.2. Draft Equipment and Animals

Households in the northern research areas were more likely to be fully equipped
and also owned more equipment per adult equivalent producer and per household
(Table 3-5). This is to some extent a legacy of the Wm; M an equipment
distribution credit program in effect from 1960 to 1980, which was operative primarily in
the Peanut Basin and the Lower Casamance (Havard, 1986a; see Ndoye, 1980 for de-

tails of the PA). The sources, means of payment, date of acquisition and condition of

TABLE 3-5: DRAFT EQUIPMENT AND ANIMAL OWNERSHIP
(Standard Error)

 

 

 

 

Item North South Total
Equipment Non-Usersg 0% 22% 13%
Borrowers 13% 26% 21%
Owner/Borrowers 15% 24% 20%
Fully Equipped 72%. 28% 46%
Mean FCFAb ‘13ng
WHO :
Equipment 25414 11541 17216
(2262) (1265 (1282)
Draft Animalsc 18151 9269 12903
b, (1765) (1413) (1144)
per household 43565 20809 30119
(3729) (2470 (2251)
per AEP 8609 4395 6119
(605) (509) (416)

 

Source: ISRA/MSU FSP Surveys (1986).

a. Equipment non-users neither own nor use a (draft)

hoe or a plow. Borrowers do not own but use
Owner-borrowers own

implements and draft animals.
either the animal or the implement, but not both;
they use both by borrowing either the animal or

the implement. Fully equipped households own

both a hoe and/or a plow and a horse and/or oxen.
b. FCFA is the West African Currency Unit (50 FCFA -
1 French Franc; 300 FCFA - approximately 1 USS).
c. Equipment and draft animal (amortized) prices are
from SODEFITEX and field surveys, respectively (see

Appendix A-l).

71

the equipment are described in GH (April 1987).4 Recent investments in draft equip-

ment (in 1987) are described in GR (May 1988).
3.1.3. Chemical Products

Less than one-half of the farm households studied reported using mineral fer-
tilizer in the period 1985-87 (Table 3-6), and 20% are estimated to have received 75 % of
the fertilizer distributed in 1986 (GH, April 1988, p.7).5 In the south, 28.6% (4/14) of
the households using herbicide do not use equipment, suggesting they may rely on

herbicides as a substitute for mechanical weeding At the same time, only 20% (4/20) of

the households not using equipment in the south used herbicides.

TABLE 3-5: USE OF CHEHICAL PRODUCTS

(Standard Errors)

 

 

 

Product/Item North South Total
WW9:
Herbicides 2%. 23% 15%
Insecticides 15% 38% 29%
Fungicides 57% 4% 26%
Fertilizer (NPK/Urea) 49% 46% 47%
MW
Kgs. of N-P-K 69 91 82
(12) (14) (10)
Kgs. of Urea 20 20 20
a (5) (4) (3)
per Household 7186 9748 8958
(1297) (1522) (1044)
per AEP 1560 2135 1900
(255) (315) (214)

 

Sggngg: ISRA/MSU FSP Surveys (1986).

a. Price used is 88 FCFA/kg for both N-P-K and urea.

 

4See also section 3.1.5.

5In the southern research areas, the estimate of the proportion of farmers using
improved inputs tends to be upward-biased due to a small over-sampling of farmers working

with SODEFITEX.

72
Table 3-7 shows the average distribution of chemical products among different
members of the household. The relatively large quantity of mineral fertilizer used by
married sons is likely to be related to their propensity to grow cotton (especially in the
south). GH (May 1987 and April 1988) describe the use of chemical products on

different crops in recent years.

TABLE 3-7: DEPENDENCY STATUS AND CHEHICAL PRODUCT USAGE.a

 

 

 

Dependents
Chemical Household Female Married Other Male Non-Fam. thds.
Product Head (Wife) Son Relative Labor Nbr. %
N-P-K (kg) 119.7 12.3 46.7 18.2 0.0 89 44
Urea (kg) 64.0 3.3 13.0 5.7 0.0 51 26
Insect. (kg) 4.9 0.8 3.4 2.1 0.0 59 29
Herbicide (l) 3.6 0.3 1.0 0.4 0.0 30 15
Fungicide (kg) 0.5 0.0 0.0 0.7 0.0 49 26

 

Souroo: ISRA/MSU FSP Surveys, 1986/7
a. Data are per adult-producer-equivalent.

The relative importance of public and private sources of fertilizer is discussed in
Section 3.1.5. Section 3.2. discusses farmer perceptions of mineral fertilizer, their
anticipated reaction to USAID’S Agricultural Production Support (APS) project, and
general issues involved in privatizing input distribution, while Section 3.3. examines
factors affecting the use of mineral fertilizer at the household level.

.3Jsk Sheds

Millet and maize are of equal importance among coarse grains seeded in the
north, while maize dominates sorghum in the southern areas (Table 3-8). The large
quantity of peanuts seeded in the north relative to the south is noteworthy. Higher
precipitation in the south offers more opportunities for cotton cultivation as an alterna-

tive to peanuts. Heads of horizontally and/ or vertically integrated households were on

TABLE 3-8: HOUSEHOLD seen usaara
(Standard Errors)

 

 

 

Kilograms and Valueb North South Total
Millet 13.9 2.2 7.0
(1.6) (.50) (.82)

Sorghum 3.2 13.6 9.3
(.75) (1.6) (1.0)

Maize 13.9 18.2 16.4
(1 5) (1.8) (1.2)

Total Coarse Grains 31.0 34.0 32.7
(2.6) (2.8) (2.0)

Per AEP 6.3 8.7 7.7
(.52) (.71) (.47)

Rice 0.1 31.3 18.5
(.05) (4.5) (2.8)

Total Cereals (FCFA) 3101 7471 5683
(262) (697)_ (451)

Cereals (FCFA)/AEP 634 1731 1282
(51) (119) (83)

Peanuts (unshelled) 590 136 322
(61) (19) (32)

Cotton 7.3 33.7 22.9
(2.4) (62) (3.6)

Total Cash Crop (FCFA) 71681 20321 41332
(7351) (2407) (3772)

Cash Crop (FCFA)/AEP 13148 4534 8058
(999) (491) (584)

 

Souroo: ISRA/MSU FSP Surveys (1986).

a. See Chapter IV for a description of crop mixes.

b. Prices used in calculations are: Coarse grains-100
FCFA/kg (observed market price at planting in 1986);
rice-130 FCFA/kg; peanuts-120 FCFA/kg;

cotton-120 FCFA/kg.

74
average able to muster larger total quantities of peanut seed, enabling them to provide
members of their household with a similar amount of seeds as did farmers heading
single families (Table 3-9). Households heads hiring non-family workers not only laid
claim to more total peanut seeds, but in the north on average also offered sigrificantly
(at the 5% level) more seeds per worker.

In the southern areas these relationships are not as straightforward, but the
results generally suggest that household heads who attract additional workers acquire
more peanut seeds.6 The relationship between household organization and labor use
and coarse gain seeds is examined in detail in Section 3.3.

Table 3-9 illustrates the importance of SONACOS in providing peanut seed.
This raises the question, if SONACOS discontinues its peanut seed credit program, and
no viable alternative system is developed, will household heads be able to carry-over
enough of their own seed to attract additional workers? Cotton is a poor substitute for
peanuts in the drier climate of the Peanut Basin, and coarse gains are not likely to
become a viable alternative unless food markets become more profitable, reliable and
predictable (see also the discussion in Ch. IV on this subject).

3.1.5. Factors Associated with Equipment Ownership

Given the apparent importance, demonstrated in the next chapter, of agricultural

draft equipment in raising crop production levels of households, Table 3-10 shows select

 

6Peanut seed stocks are difficult to maintain since the seed is expensive (relative to low
cash balances of households at planting); seed requirements per hectare are high (ca. 120
kgs/ha); and the peanut seed-output conversion ratio is low (approximately 1:10 in the
Peanut Basin, see Gaye, 1987). In contrast, cotton has a low per hectare seed ratio and is
distributed gatis as part of the input package provided by SODEFITEX.

75

TABLE 3-9: CASH CROP SEEDS (KGS) AS LABOR INCENTIVES

(Standard Errors)

 

Household
Organization
and Labor Use

 

 

 

Single Family
Horizontal Int.
Vertical Integ.
Both (HI*VI)

Use Non-Fam. Lab.

Family Lab. only

With Parastatala
Not with Parast.

North South
Peanut Seed Cotton Seed Peanut Seed Cotton Seed
Total /Vkr. Total /Hkr. Total /Hkr. Total /Wkr.
-------------- Kilograms of Seed Planted-------------
461 116 5.5 L 4 101 28 19 4.8
(54) (13) (2. 3) (. 370) (16) (4.1) (5.3) (1.2)
562 101 96 33 154 41 23 8.1
(175) (17) (6. 9) (2. 1) (46) (13) (7.8) (2.7)
786 96 205 31 83 12.8
(145) (17) (6. 6) (1.4) (88) (14) (23) (3.5)
995 106 178 23 55 6.5
(229) (27) (7. O) (1.4) (51) (6.6) (24) (2.7)
880 134 4. 9 1.2 172 32 30 7.2
(101) (14) (2. 4) (.72) (69) (8.5) (17) (3.8)
334 84 25 131 31 34 6.8
(47) (9) (4.1) (1.1) (20) (4.5) (6.1) (1.1)
719 123 39.4 10.3 196 41 93 19.0
(96) (12) (9.7) (2.7) (31) (5.9) (11) (1.9)
436 90 0.0 0.0 61 18 0.0
(64) (10) (0.0) (0.0) (15) (5.1) (1. 5O) L O)

 

Souroo: ISRA/MSU FSP Surveys, 1986/7
a. Parastatal is SONACOS for peanut seed, SODEFITEX for cotton.

76

TABLE 3-10: VARIABLES ASSOCIATED NITH EQUIPMENT OWNERSHIP

 

Equipment Value per Horker (FCFA)

 

 

North South Total
Variable Mean StErr Mean StErr Mean StErr
Age of Household
Head: 20-39 yrs 23151 3258 11070 2200 16104 2007
40-54 yrs 26595 4773 9928 1711 16056 2264
55-90 yrs 26286 3730 13742 2645 19297 2321
Nbr. of Hives: 0 9347 5809 4788 4788 7275 3726
1 19355 2319 8618 1562 12723 1403
2 33571 3367 15810 2380 23704 2242
Non-Fulani 26438 2834 7605 1946 20076 2237
Fulani 23845 3775 12610 1505 15510 1538
Nuclear Family 19052 1861 9427 1591 13107 1294
Horizont. Integ. 20925 5040 8544 2379 14015 2728
Vertically Inte. 38169 6116 17662 3761 27916 3940
Both (HI*VI) 46172 6535 19448 4325 27308 4635
Fan. Lab. Only 18246 2461 10693 1299 12903 1197
Use Nonfam. Lab. 33525 3530 18318 4366 29649 2986
Land-Constraint 36881 4488 14693 7428 34734 4262
Not a constraint 19356 2113 11458 1288 13964 1138
'85 Stcks < 9 mth 17463 3041 8728 1368 10821 1318
> 9 months 28567 2829 16207 2340 23235 1986
Cattle Owner 33598 5257 16932 1992 22281 2313
Non-Owner 18856 2923 5947 1126 10197 1387
Not w/ SODEFITEX 26300 2724 9363 1509 17050 1655
Hith SODEFITEX 23038 4058 14902 2148 17534 1996
Not w/ SODEVA 24038 2383 11541 1265 15960 1252
Hith SODEVA 30594 5970 . . 30594 5970
Not w/ SONACOS 24966 3221 7819 1784 14947 1911
Hith SONACOS 25790 3197 14518 1698 19068 1716
No Off-Farm Inc. 30331 4219 13459 2666 21604 2681
Off-Farm Income 22816 2603 10879 1434 15398 1411
No Borrowing 25232 2900 12640 1538 17057 1512
Borrowed Cash 25665 3653 8353 2037 17550 2407

 

(continued)

77
TABLE 3-10: Continued

 

Equipment Value per Horker (FCFA)

 

 

North South Total
Variable Mean StErr Mean StErr Mean StErr
Age of Household
Head: 20-39 yrs 6260 872 3192 611 4471 541
40-54 yrs 4599 535 2152 357 3052 330
55-90 yrs 4200 499 2091 414 3025 341
Nbr. of Hives: 0 3713 2146 1127 1127 2537 1286
1 5185 631 2337 399 3426 370
2 4946 383 2740 403 3721 308
Non-Fulani 5042 493 1576 388 3871 399
Fulani 4831 596 2679 317 3234 292
Nuclear Family 5450 646 2366 373 3545 367
Horizont. Integ. 4116 733 2383 632 3149 491
Vertically Inte. 4783 555 2564 512 3674 417
Both (HI*VI) 4965 470 2787 914 3428 695
Fam. Lab. Only 4865 547 2280 280 3036 272
Use Nonfam. Lab. 5065 521 3748 768 4729 439
Land-Constraint 5559. 465 2250 1359 5239 468
Not a constraint 4642 520 2448 271 3144 259
’85 Stcks < 9 mt 3583 522 1770 298 2204 269
> 9 months 5504 468 3560 461 4665 344
Cattle Owner 5925 848 3423 407 4226 406
Non-Owner 3980 565 1389 265 2242 289
Not w/ SODEFITEX 5171 458 2011 319 3445 304
Hith SODEFITEX 4390 652 3110 448 3524 374
Not w/ SODEVA 4586 370 2443 265 3201 228
Hith SODEVA 6363 1102 . . 6363 1102
Not w/ SONACOS 5521 666 1848 435 3375 420
Hith SONACOS 4486 405 2919 318 3552 260
No Off-Farm Income 5749 570 3328 580 4497 434
Off-Farm Income 4541 486 2138 291 3048 275
No Borrowing 5103 460 2752 328 3576 283
Borrowed Cash 4760 643 1548 371 3254 431

 

Sourog: ISRA/MSU FSP Surveys, 1986/7

78
variables hypothesized7 to be associated with different levels of equipment ownership,
without necessarily implying causality between the variables. Households headed by
younger individualsg; facing a land constraint; having food stocks out of 1985 production
which lasted 9 months or longerg; and households owning cattle had more equipment at
their disposaL regardless of where they resided (i.e., north or south).

In contrast, households which pursued an ofi-farm activity, or had borrowed cash
during the 1986 hungy season, owned less equipment. This partial analysis, and this
data set, therefore suggest that households do not necessarily have to pursue off-farm
activities to maintain equipment ownership over time. Participation in SODEFITEX’s
crop production progam made a noticeable difference only in the south; this reflects
more recent sales (of ex-ONCAD equipment stocks) by SODEFITEX in the Casa-
mance, and the effects of the Program Agricolo in the northern research areas. The
receipt of peanut seed from SONACOS seems to facilitate equipment ownership in the
southlo, but not in the north.

To examine more precisely the role of parastatals in facilitating equipment
ownership, Table 3-1 1 shows the status of households with respect to equipment owner-
ship, borrowing and non-use as it relates to participation in parastatal The table
provides some support for the assertion that parastatals make a difference in terms of

which households own equipment, but there are exceptions to this rule. It is of course

 

7These variables are demogaphic characteristics of the household and entitlements,
which in Chapter II were argued to affect the performance (including equipment ownership)
of the household.

8h is not clear that this results from increased risk-taking behavior of younger males.
It is also conceivable that some of the young males in the sample had just ”inherited” the
equipment of the household.

9Higher food stocks may have been made possible by equipment ownership, and at the
same time made equipment ownership possible by obviating forced equipment sales, as may
have occurred in other households.

10Again, it is not possible to establish causality here.

79

TABLE 3-11: EQUIPMENT STATUSa AS A FUNCTION OF PARASTATAL
AFFILIATION OF THE HOUSEHOLD
Percent of households responding

Parastatal North uSouth
Affiliation 80 DB FE NU BO 08 FE

 

 

None 43% 25% 21% 38% 48% 13% 22%
SODEFITEXb 33% 23% 27% 27% 23% 52% 51%
SONACOS 44% 69% 53% 27% 45% 66% 74%
SODEVA 11% 8% 25% - - - - - - - -

Sample Size:c 9 13 59 22 31 29 35
SONACOS**a 33x 60% 47% 25% 33% 71% 55%
Sample Size: 6 10 43 16 24 14 17

Souroo: ISRA/MSU FSP Surveys, 1986/7

a. NU-equipment non-user; BO-borrower (non-owner);
OB-owner/borrow (partial owner); FE-fully equipped.

b. Maize and/or cotton programs.

c. For the sample working with a parastatal.

d. For the sample not working with SODEFITEX programs.

 

 

plausible that some of the fully equipped households worked with one or more of the
parastatals in the past, and dropped their affiliation once the equipment had been paid
off. Another important feature of equipment ownership is that fully equipped house-
holds on average have significantly larger labor forces than borrowing and non-using
households (i.e., 5.7 vs. 4.5, 3.0 and 3.0 adult-equivalent producers in fully equipped,
owner/borrowing, borrowing and non-using households, respectively. This may suggest a
minimum size labor force is necessary to profitably use equipment; at the same time, it
may be easier to attract other workers (family and non-family) the larger the stock of
equhnnent

Table 3-10 also showed that the total value of equipment owned rises dramati-
cally from nuclear families to horizontally (HI), vertically (VI) and horizontally-and-
vertically (HI‘VI) integated households in the northern areas. The same relationship

80

does not hold on a per worker basis however, _or in the southern areas, except that
households that are integated in both senses in the south on average own the largest
amount of total equipment. A hypothesis raised earlier was that the sharing of equip-
ment constituted one of the benefits of being a member of a oonooaion (i.e., horizontal
integation). Table 3- 12 provides some (weak) support for this hypothesis, with horizon-
tally integated households showing the largest proportion of borrowers for each of the
equipment types and animals listed in 5 out of 8 cases in the north (with one tie of 11%
for plows) and 3 out of 7 cases in the south (peanut plant lifters are not owned in the
south). In the north, 47.4% of the horizontally integated households are equipment
and/or draft animal borrowers, as compared with 25.6%, 11.2% and 20% of the nuclear
families, vertically integated and HI‘VI households, respectively. In the southern area
these percentages are 45.8%, 58.7%, 55.6% and 16.7%. Table 3-12 also shows that
HI’VI households on average had the largest labor force of all households. Finally, in
the north 58% of the HI households work with parastatals, as compared to 21%, 28%
and 20% of the nuclear, VI and HI‘VI households.

3.1.6. Private versus Public Involvement in Input Markets

Parastatals (especially SODEFITEX) were the primary source of chemical and
cash crop inputs for farmers in 1985-87, with the private sector taking on a
limited role even in 1987, when it is estimated to have sold only 4% of the total
fertilizer used (GH, April 1988).11 Improved inputs are sold on a credit basis by
parastatals. Given the predominance of parastatals in the marketing of technical inputs,
it would be useful to examine parastatal’s prowdures for identifying farmers with whom
they work, and to analyze their (negative and positive) experiences with credit ganted to
different kinds of farmers.

 

11The participation of households in parastatal crop production progams is described
in Diagana and Goetz (December, 1987).

81

 

 

 

TABLE 3-12: EQUIPMENT BORROHING AND SIZE OF LABOR FORCE
ACCORDING TO HOUSEHOLD ORGANIZATION AND REGION
(Standard Errors)

North South
Equipment Nucl. VI HI HI*VI Nucl. VI HI HI*V1
---------- Percent of Households per Stratum----------
Hoes 15 6 26 O 21 17 25 8
(6) (6) (10) (5) (9) (9) (8)
Plows 3 11 11 O 21 17 17 O
(3) (8) (7) (5) (9) (8)
Soule-b 15 6 16 0 Not Owned
veuse (6) (6) (9)
Seeders 15 11 32 0 3O 6 33 17
(6) (8) ll (6) (6) (10) (11)
Cart 46 22 21 O 41 33 46 8
(8) (10) (10) (6) (11) (10) (8)
Oxen 3 6 11 0 22 39 33 8
(3) (6) (7) (5) (12) (10) (8)
Horse 10 O 21 O 5 O 4 0
(5) (10) (3) (4)
Donkey 3 6 5 O 14 22 21 O
(3) (6) (5) (4) (10) (8)
Horkers 4.1 7.9 4.4 9.8 3.6 6.6 3.5 7.9
(AEPs) (.3) (.8) (.6) (1.8) (.2) (.5) (.3) (.7)
Valid N 39 18 19 5 63 18 24 12

 

Souroo: ISRA/MSU FSP Surveys, 1986/7
a. Nucl.-nuclear household, VI-vertically, HI-horizontally
integrated.

b. A peanut plant-lifting implement designed to facilitate the
harvesting of peanuts.

82

In the scant instances where credit is ganted by private traders, it tends to be
used mainly for consumption purposes (GH, May 1987). Gaye (1987) argues that the
historical prominence of SONACOS in the peanut credit market has broken the produc-
tion credit ties sometimes observed between traders and producers in other countries.
Farmers sampled did purchase fungicides, draft equipment, draft animals and peanut
seed from private (non-cereal) traders during the 1987 dry season, suggesting a wil-
lingness of some individuals to handle certain agicultural inputs (the durables may have
represented former collateral from other transactions).

As of 1986, about one-half of the agicultural equipment (hoes, plows and
seeders) owned by farmers had been obtained from a parastatal (on credit), while other
producers supplied implements in roughly one-third of the cases. These latter transac-
tions largely reflect distress sales of (used) equipment, since farmers are not known to
manufacture equipment.

Most cereals seed were obtained from carry-over stocks, cotton seed from
SODEFITEX, and peanuts were obtained from SONACOS as well as carry-over stocks.
Given these results on the ”manifest behavior" of farmers, we are now ready to turn to
their “potential behavior“. After reviewing farmers’ opinions and anticipated responses
to input market reforms in the next section, section 3.2.6. discusses issues involved in
the privatization of input delivery.

3.2. INVESTMENT PRIORITIES, EXPECTATIONS AND THE NEW INPUT

DISTRIBUTION POLICY

This section presents results on what farmers say they would have done in the
1987 season if they had had more cash; what levels of returns they expect from invest-
ments in their crops; and their perceptions and knowledge of mineral fertilizer use. The
question of whether or not they have better investment alternatives than using fertilizer

on maize is examined, and their opinions of the new input distribution policy are

83

discussed. Asking farmers how they feel about new technologies and market reforms
(Chapter V), and why, is a useful starting point in providing feedback to policy makers
contemplating progams designed to increase the productivity of producers.

3.2.1. Investment PrioritiesattheBeglnningofthe 1987 Season

To determine their investment/cash expenditure priorities at the beginning of the
1987 agricultural season, household heads were asked how they would have spent 15,000
FCFA (see notes to Table 3-13 for the formulation of the question), such as may have
been obtained in kind through the APS. Two follow-up questions enquired how the
funds would have been used as second and third priorities if the respective previous
need had already been met (Table 3-13).12

Table 3- 13 shows that farmers would have used credit to buy food as an irnpor-
tant first priority-following a not-so-bad cereals crop in 1986-as well as peanut seed,
agicultural equipment or labor. Gaye (1986) argues peanut price policy changes during
1984 and 1985 translated into a perceived-and guaranteed—price increase of 80% for
farmers (from 50 to 90 FCFA/kg) due to changes in the credit retention procedure; in
comparison, announced coarse gain floor prices were raised from 60 to 70 FCFA/kg in
the same period (a 17% change); this may explain the strong preference for peanut seed.

As indicated earlier (and tested in Section 3.3.), Kelly and Gaye (1986) argue
that to attract seasonal labor, and to retain sons in the household, household heads need
to provide these workers with cash crop opportunities.13 Most household heads (99%)
surveyed in this study believe it is their responsibility to supply other household mem-
bers with peanut seeds. Consequently, those citing peanut purchases as a first priority
probably not only perceive this crop to be more profitable, but would also use the seed

 

12This question was originally addressed to farmers in the Peanut Basin by Kelly and
Gaye (1986).

13See also the discussion in Kelly (1988c).

TABLE 3-13: FARMER PRIORITY USES OF 15,000 FCFA;a
BEGINNING OF 1987 SEASON
Number and Percent of Households Reporting

 

 

 

 

North South Total

First Priority

Buy Food ............... 41 47% 68 55% 109 52%
Buy Peanut Seed ........ 27 31% 17 14% 44 21%
Buy/Repair Equipment... 14 16% 10 8% 24 11%
Hire Agr. Labor ........ O 0% 19 15% 19 9%
Buy Fgrtilizer ......... 3 3% 1 1% 4 2%
Other ................ 2 2% 9 8% 11 5%
Second Priority

Buy Peanut Seed ........ 33 38% 17 14% 50 24%
Buy Food ............... 12 14% 28 23% 4O 19%
Buy/Repair Equipment... 17 20% 4 3% 21 10%
Hire Agr. Labor ........ l 1%, 20 16% 21 10%
Buy Fsrtilizer ......... 11 13% 3 2%. 14 7%
Other ................ 23 27% 53 43% 64 31%
Third Priority

Buy/Repair Equipment... 17 20% 9 7% 26 12%
Buy Food ............... 12 14% 11 9% 23 11%
Buy Fertilizer ......... 15 17% 6 5% 21 10%
Buy Peanut Seed ........ 2 2%. 17 14% 19 9%
Hire ggr. Labor ........ 4 5% 14 11% 18 9%
Other ................ 37 41% 66 54% 103 49%

 

Souroo: ISRA/MSU FSP Surveys, 1986/7

a. Following Kelly and Gaye (1986) farmers were asked how they
would have invested an additional 15,000 FCFA prior to the
beginning of the rainy season (open-ended question format).
Two follow-up questions asked what they would have done
with 15,000 FCFA if the previously cited need had already
been met. It is conceivable that equipment purchases were
rarely cited because most pieces cost more than 15,000
FCFA. The same may be true of large ruminants.

b. Includes save the money; other agricultural and non-
agricultural use; purchase/trade in livestock: purchase of
draft animals.

85
to attract additional labor to the household (see also the discussion in 4.1.5.). The same
may be true for those who would have bought food, since this serves as a partial pay-
ment for seasonal workers. 14
While fertilizer purchases (in retrospect) did not rank highly among stated
priority uses of funds at the beginning of the 1987 rainy season, 44% of the household
heads who had not received fertilizer as of June 1987 indicated the that they planned to

buy it if it were available on credit (Table 3-14).15

TABLE 3-14: PURCHASE INTENTIONS FOR FARMERS NOT HAVING
RECEIVED FERTILIZER AS OF JUNE 1987, BY REGION
Number and Percent of Households Responding

 

 

 

 

 

Do Not Plan Plan to Buy Plan to Buy Not Yet

To Buy if on Credit Credit or Cash Decided

Region N % N % N % N %
North 28 36% 38 49% 12 15% O 0%
South 28 30% 37 40% 9 10% 19 20%
Total 56 33% 75 44% 21 12% 19 11%

 

Souroo: ISRA/MSU FSP Surveys, 1986/7
Note: Data for the 80% of households in the sample not having
received fertilizer as of June, 1987.

Since many household heads responded that they would have bought food with
additional funds at the beginning of the 1987 season, there may be a tendency to resell
fertilizer obtained on credit to purchase food. The decision to resell improved inputs

received through the APS may depend on two factors: a) the urgency to resell, which will

 

14In fact, it is not possible to attract seasonal farm labor unless one can provide food
during the gowing season.

15The final survey in September/ October 1987 shows less than half of the households
sampled acquired fertilizer in 1987 (most of it obtained on credit). The results suggest most
farmers will buy fertilizer on credit but not use [their own] cash to buy it. This may be
related to the belief that if crops receiving fertilizer fail (eg. due to drought) someone else
should share the loss with the farmer (see also section 5.1.6).

86
be higher for household heads with a precarious food situation and b) relative prices of
the traded goods. For example, farmers re-selling fertilizer to buy peanut seed will put
downward pressure on fertilizer prices until it is no longer profitable to sell it in
exchange for peanuts. This will happen sooner the larger the quantity of improved
inputs that becomes available in a given area.

In the southern area a large number of respondents would have m additional
funds, if their first need had already been met. The following section offers hints as to
why these farmers may have preferred not to commit further funds to agiculture.

3.2.2. Expectations for Returns on Crop Investments

To develop an idea of their expected goss benefit-cost ratios, interviewees were
asked what level of returns they would expect if they invested 5,000 and 20,000 FCFA in
an agricultural crop. The question initially confounded farmers, who tend to believe
returns to investments depend on Allah and are not for mortals to guess at. Another
problem arose from the fact that some farmers anticipated if they were to invest $000
FCFA, e.g., for additional manual weeding in their peanut fields, an increase in yields
would raise labor costs for harvesting and transporting. Since it was not possible to
control for this effect, some of the data in Table 3-15 reflect "true" goss returns, taking
into account secondary costs associated with the investment, while others exclude such
costs.

The goss benefit/cost ratio distribution ranges from 1.16 to 8.00 for the 5,000
FCFA investment, and exhibits both a mode and median of 2.00. Two key insights may
be gained from Table 3-15:

a. Farmers in the south are on average less optimistic about returns from agicul-
tural crop activities, despite the alleged high potential of this area.
b. Farmers in both areas on average report diminishing marginal returns to capital

investments.

87

TABLE 3-15: FCFA RETURNS EXPECTED BY FARMERS FOR THO
INVESTMENTS IN AN AGRICULTURAL CROP ACTIVITY
(Standard Error)

 

Net Return 0 5000 Net return 0 20000 Gross Benefit-

 

 

 

Cost Ratio
Mean Range Mean Range @5000 020000
North 8194 800-35000 27286 3200-140000 2.64 2.34
(1039) (3301)
South 5429 1000-25000 18945 1700-50000 2.09 1.95
(365) (1064)
Total 6351 800-35000 21726 3200-140000 2.27 2.09
(434) (1340)

 

Sourog: ISRA/MSU FSP Surveys, 1986/7

Note: 52/211 household heads were not able or willing to advance a
return. Both the mode and the median are 5000 FCFA for the 5000
FCFA investment (both regions); the mode is 20000 FCFA for the
20000 FCFA investment in both regions; the median is 20000 FCFA in
the north, 18000 FCFA in the south.

In the northern areas, fully-equipped farmers on average reported a higher
expected return for the 5,000 FCFA investment (8,766 FCFA). In contrast, fully-
equipped southern farmers expected a 15 % lower return (4,625 FCFA) than the average.
A plausible explanation is as follows. When asked toward the end of the 1987 gowing
season which crop they would have gown more of if they had been able to, 83% of the
southern farmers responded with coarse gains; half of the northern farmers responded
with peanuts. If farmers were thinking of these crops when responding to the investment
question (posed at the beginning of the season), this would explain why northern farmers
were on average more optimistic about expected returns. For both areas, those who
would have gown more peanuts in 1987 reported a higher expected goss benefit (2.96)
than those who would have expanded into cereals production (2.03).16

 

16The null-hypothesis of equality of the two means is rejected at the 1% level of
sigrificance.

88
3.2.3 Farmers’ Knowledge and Perceptions of Fertilizer
Most farm households in the research areas have used fertilizer at some point in
the past. 17 Nevertheless, many respondents did not believe they were better able to
assess the quantitative fertilizer needs of their cereals fields than an extension agent
(Table 3-16). Farmers participating in SODEFTTEX’S cotton and / or maize progam in

1986 were more likely to answer this question affirrnatively than those who were not.

TABLE 3-16: RESPONSES TO THE QUESTION '00 YOU BELIEVE YOU
KNOH THE QUANTITATIVE FERTILIZER NEEDS OF YOUR CEREALS
FIELDS BETTER THAN AN EXTENSION AGENT?‘I
Number and Percent of Households Responding

 

 

Response North South Total
No 44 52% 71 58% 115 56%
Yes 39 46% 50 41% 89 43%
Don ’ t Know 2 2% 1 1% 3 1%

 

Sourgo: ISRA/MSU FSP Surveys, 1986/7

Agonomists, including SODEFITEX researchers in Tambacounda, recommend
two separate applications of fertilizer to maize: prior to land preparation and after the
first weeding.18 In contrast, Figure 3-2 shows most farmers would apply fertilizer only
once, after the first weeding, when the plant population is well-established. This
confirms the existence of a rule of thumb alluded to by Kelly and Gaye (1986), that far-
mers tend not to apply fertilizer unless rains are satisfactory and they have "seen the
plant" gow. Farmers working with SODEFITEX in the south (1986), were more likely

to respond they would apply fertilizer after tne first weeding than those in the south not

 

l7See Goetz and Holtzrnan (May, 1987).

18See, e.g., Martin (1988). SODEFITEX information was obtained through personal
communication (1987). SODEFITEX does not recommend use of fertilizer on (unimproved)
millet and sorghum, since this was found to be uneconomical.

89

FIGURE 3-2: TIME IN THE CROP CYCLE
AT HHICH FARMERS HOULD APPLY FERTILIZER

 

 

 

 

 

 

Application Count Percent
Before Seeding 36 — 17%
After Seeding 29 — 14%
After 1511 "9991119 142 — 58%
Other 2 . 1%
l l l !
O 40 80 120 160
Frequency

 

Sourog: ISRA/MSU FSP Surveys, 1986/7

working with this parastataL Furthermore, southern farmers more frequently (72 vs.
62%) responded with this date of application than northern farmers, suggesting many
worry about stronger weed gowth due to early fertilizer use (this was confirmed in
informal interviews).

It is plausible that the strategy of applying fertilizer only after plant emergence
has a higher average pay-off than applying fertilizer twice (as on experiment stations),
given the uncertainty of rains in Senegal. The results also suggest provisions should be
made for distributing weed-control equipment and chemicals along with the fertilizer
under the APS. Surveys of village chiefs in the areas reveal they are generally aware of
the benefits (and potential toxic efi'ects on humans and livestock) of using herbicides;
their experience with this chemical stems mainly from cash crops (cotton). We have
little information on the perceived profitability of herbicides applied to cereals, since it is
not widely used. However, of 12 farmers applying herbicide to maize in 1985 or 1986, 11
were convinced its use is "worthwhile'.19 A partial budget in Appendix A-2 shows some

 

19All southern farmers using it were convinced herbicide use was worthwhile. For the
overall sample, 11 out of 17 northern and 82 out of 121 southern farmers claiming to be
able to assess the benefit of herbicide relative to its cost reported it to be "worthwhile”.

90
of the costs and benefits involved in herbicide use on maize for one representative
SODEFITEX farmer.

One question for Senegal’s NAP is how farmers perceive of mineral as compared
to organic fertilizers (manure) when applied to cereals.20 Farmers who are strongly
convinced manure is superior (Table 3-17) may convert mineral fertilizer received on
credit into livestock holdings. In this case full benefits would accrue only after a period
of years, making it more difficult to repay a one-season production credit. As will
become evident below, not all the responses shown in Table 3-17 refer to the agonomic

efficacy of the two types of fertilizers.

TABLE 3-17: FARMERS’ PERCEPTIONS ABOUT THE RELATIVE EF§ICACY
OF MANURE AND MINERAL FERTILIZER APPLIED TO MAIZE
Number and Percent of Households

 

 

Perception North South Total
No difference 10 12%. 4 3% 14 7%
Manure better 33 41% 37 30% 7O 34%
Fert. better 38 47% 82 67% 120 59%

Total 81 100% 123 100% 204 100%

 

Souroo: ISRA/MSU FSP Surveys, 1986/7
a. Similar results were obtained for millet and sorghum.

Regional averages tend to mask important differences among triads. In N doga,
for example, most farmers believed manure is "better" for cereals, while in Diega more
than 80% stated mineral fertilizer is better. It is plausible that perceptions depend on
how effectively manure and fertilizer interact with the predominant soils of a given area,
suggesting more site-specific agonomic research is needed to improve the effectiveness

of fertilizer (Kelly, 1988a, reports widely different responses to fertilizer, depending on

 

20A survey of village chiefs showed the practice of applying manure and mineral
fertilizer together is rare. See also note c in Table 3-18, however.

91
the area). Nevertheless, the larger proportion of southern farmers favoring mineral
fertilizer is likely to reflect the beneficial interaction between fertilizer and higher
rainfall. Table 3-18 shows responses as to why farmers believe mineral or organic fertili-
zer is more effective.

Using a base of 10 sacks of cereals harvested in a field without fertilizer and
animal manure under 1987 rainfall conditions, household heads were then asked how
many sacks they would expect to harvest if they had applied "sufficient" amounts of
mineral fertilizer (see Table 3-19). Parallel questions covered situations of poor and
abundant rainfall.21

Virtually all farmers believe mineral fertilizer can augnent yields of cereals.22
For 1987, the average expected response was highest for field maize in the southern, and
highest for sorghum-not maize-in the northern area. Lower expectations for maize may
reflect temporary drought conditions during the 1987 gowing season. In both areas
expected yield averages for home maize were consistently higher than for field maize,
presumably reflecting the higher organic matter content of soils near the home.

A cross-tabulation of average expected maize yields under 1987 rainfall condi-
tions with fertilizer revealed similar results regardless of whether or not the farmer
believed he could better assess the quantitative fertilizer needs of his cereals fields, and
whether or not he believed manure was superior to mineral fertilizer. Farmers who had
participated in SODEFITEX’S maize progam in 1986 on average advanced yield expect-

ations similar to those who had not (2% lower in the north, 4% lower in the south),

 

21Some farmers indicated expected yields would not change under conditions of
abundant rainfall as in the distant past, for example for millet, since it does not demand
much moisture. Others advanced identical yield expectations for the abundant and 1987
rainfall scenario since 'too much rain washes away the fertilizer".

22’I‘wo farmers in Ndoga pose an exception, responding they would only be able to
increase yields with higher rainfall; this may be related to the drought during the 1987
gowing season.

TABLE 3-18: REASONS GIVEN BY FARMERS FOR PREFERR G ORGANIC
OR MINERAL FERTILIZER, APPLIED TO MAIZE
Number and Percent of Households Responding

 

 

 

 

ORGANIC FERTILIZER BETTER N S T %
1. Manure lasts in the soil for multiple years... 17 14 31 46%
2. Increases yields more than fertilizer ........ 13 14 27 40%
3. Cattls urine acts as an herbicide ............ O 5 5 7%
4. Other ...................................... 3 l 4 6%

Total: 33 34 6—7100%
MINERAL FERTILIZER BETTER N S T %
1. Increases yields more than manure (stronger
plants and larger ears) ...................... 14 50 64 53%
2. Fixes roots solidly in the ground ............ O 17 17 14%
3. Spreading is easier (manure difficult to
distribute evenly over the field) ............. 6 8 14 12%
4. Helps crops during droughts .................. 9 O 9 8%
5. Less herbs than with manure (parkage) and/or
fertilizer kills weeds (Silo/striga) ......... 6 3 9 8%
6. Immedéate result in comparison to manure ..... O 3 3 3%
7. Other ...................................... 3 1 4 4%
Total: 38 2 120 100%

 

Souroo: ISRA/MSU FSP Surveys, 1986/7

a. Similar percentages were obtained for millet and sorghum. N-North,
S-South, T-Sample, % - percent of responses by category.

b. "Other“ responses are: Fixes stems firmly in the soil; helps crop
during drought; prevents soil degradation; does not cost anything
(the latter three responses are for the North).

c. "Other” responses are: Stems turn yellow if too much manure;
Manure (cattle urine) prevents seeds from germinating; I have seen
others put fertilizer in fields which had already been manured
(parkage)--the latter three are from the North.

93

TABLE 3-19: FARMERS’ FERTILIZER YIELD RESPONSE PERCEPTIONS
(IOO-BASE YIELD HITHOUT FERTILIZER AND 1987 RAINFALL)

Millet Sorghum Field Maize
North South Total North South Total North South Total

 

 

 

 

 

 

 

1987 Rain

w/Fert.

Mean 156 191 171 166 188 183 158 194 177
Valid N 87 62 I49 37 116 153 86 95 181
Poor Rain

w/Fert.

Mean 92 92 92 45 94 86 84 84 84
Valid N 34 26 60 10 52 62 22 39 61
Poor Rain

w/o Fert.

Mean 45 49 47 29 54 49 32 42 39

Val id N 34 26 60 12 51 63 22 39 61

Souroo: ISRA/MSU FSP Surveys, 1986/7
Note: The sample size for the poor rain situation was reduced
because of enumerator illness.

 

while those who reported they would apply fertilizer early in the rainy season rather than
wait until the plant was established tended to have marginally higher yield expectations.
Finally, among farmers using fertilizer on cereals in 1987, average yield expectations
were marginally higher (4%) in the northern area, but lower in the southern area (8%).

The data therefore do not permit us to conclude that farmers not using fertilizer
in 1987 had lower expectations about the yield effect of fertilizer. In the next section
farmers’ yield expectations are used to construct goss benefit-cost ratios for using
fertilizer. These are compared to ratios listed in the APS document, as well as expected
goss benefits advanced by farmers for returns to their investments to assess the relative
profitability of investing in mineral fertilizer for cereals.

3.2.4. Fertilizer B-C Ratios Based on Farmers’ Expectations

Average goss benefit-cost ratios calculated using farmers’ cereals yield response

perceptions under 1987 rainfall conditions are shown in Appendix A-3. These estimated

94

ratios are based on official prices, official 1986 yield estimates in pertinent W
mongo, and they assume the use of traditional rather than improved seeds. Excepting
sorghum in Senegal Oriental, field maize on average yields the highest ratio, while millet
lies below 1.9 in the northern triads. If one assumes farmers cut back on fertilizer
application rates under poor rainfall conditions (by one-half), fertilizer on maize
performs better than on the other two cereals. However, if farmers do not cut back on
application rates under poor rainfall conditions, only fertilizer on millet in the Sine
Saloum remains “profitable" (Appendix A-3).

Gross benefit-cost ratios calculated using farmers’ perceptions tend to concord
well with those presented in the APS document. However, these ratios are calculated
using official prices, which are often irrelevant at the farm level. Table 3-20 illustrates

the sensitivity of the results to alternative assumptions about input/ output price ratios

for maize.

TABLE 3-20: SENSITIVITY OF BENEFIT-COST RATIOS FOR FERTILIZER USE ON
MAIZE TO ALTERNATIVE ASSUMPTIONS ABOUT INPUT-OUTPUT PRICES
(Using 1987 Rainfall conditions, Official 1986 Yields)

 

 

 

 

Sine Saloum Senegal Oriental Casamance
Assumption (1) (2) (3) (l) (2) (3) (1) (2)
FCFA/kg Px 84 109 87a 87 113 68b 90 117
FCFA/kg Py 7O 61 61 70 56 56 7O 60
P /Py 1.2 1.8 1.4 1.2 2.0 1.2 1.3 1.9
NB ( 000) 29 16 21 20 5 16 6O 39
BCR 2.4 1.6 2.0 1.9 1.2 2.0 3.5 2.3

 

Souroo: ISRA/MSU FSP Surveys, 1986/7

Px-weighted fertilizer price; P -price of maize; NB-net benefit

( OOO FCFA). BCR-benefit-cost Fatio. Note: a-20% subsidy on the
final price, b-40% subsidy (see text).

Scenario (1) is the basic situation, using official prices; (2) is a “worst-case"

situation, using average cereals prices observed in southeastern Senegal during 1987, and

95
adding a 30% mark-up to the price of fertilizer reflecting trader margins.23 In this case
the expected goss benefit cost ratio remains above 2.0 only in the Casamance. Using
these same assumptions, scenario (3) shows what prices of fertilizer delivered to farmers
in the northern areas would have to be to raise the ratio to 2.0; in terms of a subsidy on
fertilizer, this amounts to 20% and 40% on the final price charged by traders.

If one compares, finally, the distribution of benefit-cost ratios from using
fertilizer on maize, and the ratio advanced by farmers for their 20,000 FCFA investment
alternative, more southern (84%) than northern (47%) farmers would be better off using
fertilizer on maize (Figure 3-3). This may reflect both the higher profitability of fertili-
zer use in the south and a lack of competitive investment alternative. However, because
maize matures more rapidly than other crops, it is more prone to attacks by natural
predators and pests, a factor cited by many southern farmers as an impediment to
expanded field maize production (see also Ch. IV). One option may be to encourage
farmers to continue to gow maize around the compound as a "hungy season crop”, but
to delay the seeding of field maize-as a cash crop-so that it matures at the same time
as the other crops (this would also allow the farmer to assess whether or not the pattern
of rainfall is adequate to warrant the application of fertilizer). Further analysis would be
required to assess the effect of such a shift on seasonal labor constraints of the house-
hold.

 

23The 30% rate was elicited in trader surveys. It incorporates traders’ time, manage-
ment and costs associated with physical distribution (storage, transport) as well as a
premium for being held responsible for non-reirnbursements of credit by farmers; it does
not take into account the interest charged on loans to farmers, which could add another 15-
20%.

96

FIGURE 3-3: DIFFERENCES IN GROSS BENEFIT-COST RATIOS EXPECTED BY
FARMERS FOR RETURNS TO USING FERTILIZER ON MAIZE UNDER 1987
RAINFALL CONDITIONS AND THEIR OHN 20,000 FCFA CASH INVESTMENT
(y-axis-GBCR/Fertilizer on Maize minus GBCR/ZOOOO)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7
6
5
4 South
3 \
2 North ~\\;‘\
1 \\ \\\
o \\ \\\
-1 \.\ x
-2
-3 \\
-4
-5

 

O 10 20 3O 40 50 60 70 8O 90 100
Cumulative Percent of Farm Households

Souroo: ISRA/MSU FSP Surveys, 1986/7

Note: Points on the (smoothed) line shows the difference in
the return an individual farmer expects from using mineral
fertilizer on maize, both valued at official prices, and
from his own (unspecified) 20000 FCFA cash investment (see
summary statistics in Table 3-15). Hence the graph is
intended to capture the profitability of using fertilizer
relative to other investment opportunities.

97

3.2.5. Willingness to Use Selected Seeds and Fertilizer

As indicated above, there was no evidence of substantive changes in the input
distribution system in southeastern Senegal at the time of the surveys (GH, March 1988).
To anticipate farmers’ responses to the APS, hypothetical questions were used together
with information on perceptions, stated responses and observed behavior.

Essentially all household heads responded they would be interested in adopting
the fertilizer/ improved seed package to be promoted under the APS.24 However, 51 %
of the northern (=44/87) and 15 % of the southern (=18/124) household heads stated
they were unwilling to use both inputs at a cash price of 100 F /kg of fertilizer and 210
F/kg of improved seeds (the numbers drop to 42 and 11 households for credit sales in
the two regions). Responses in the northern area tend to reflect the lower price of
fertilizer available from the Gambia. Cash constraints were a particular concern in the
southern area, where many respondents reminded the interviewer they would only buy
the inputs with cash if they had sufficient funds.

Some farmers were curious to know why they were expected to pay 210
FCFA/kg for (improved) cereals seed if they only received 70 FCFA/kg for their own
production. In the case of peanut seed, Gaye (1986, p.6) reports that farmers tend to
perceive of the Government as a Wu: when it resells (indistinguishable,
improved) peanut seed to farmers at a price (110-120 FCFA/Kg) above the producer
output price (90 FCFA/Kg).25 Since cereals are a subsistence food-farmers may indeed
consume rather than plant improved seeds-and since the input/ output price difference

is threefold, it is important to explain to farmers why the improved seeds are more

 

24Some reluctance was expressed in Ndoga, where farmers have had poor experiences
with improved millet seed distributed by SODEFITEX (apparently it was particularly well-
liked by birds).

25Kelly (1988b) mentions a similar "confusion" for fertilizer which, although fabricated
in Dakar, is sold at lower (subsidized) prices in the Gambia.

98
expensive. This is critical given attitudes among some farmers that traders are exploita-
tive (see also Ch. V).
Household heads not willing to buy the inputs at ofi'icial prices, on average
proposed cash prices roughly 50% lower (Table 3-21), with insignificant regional diffe-

rences.26 These farmers are on average willing to pay a 15 % premium for the benefit of

TABLE 3-21: IMPROVED INPUT PRICES 'ACCEPTABLE' TO FARMERSa

 

 

 

 

 

 

FCFA Cash Price FCFA Credit Price
Input Mean Range N Mean Range N
Improved Seeds 102 50-200 41 119 60-200 36
Fertilizer 57 10-80 62 65 20-90 50

 

Souroo: ISRA/MSU FSP Surveys, 1986/7

a. For the subsample not willing to acquire at prices of
210 FCFA/Kg for seeds and 100 FCFA/Kg for fertilizer.
This kind of data is merely indicative and should be
interpreted with caution.

receiving the inputs on credit. Assuming a six-month loan period (J uly-December) the
premium is equivalent to a 30% annual interest rate, not unlike that paid by farmers on
private loans (GH, May 1987). Given the uncertainty of agicultural production, and the
associated credit repayment ability of producers, traders may charge similar interest
rates for loans provided to farmers under the APS.

In comparison, approximate official historical fertilizer prices and input / output
ratios are shown in Table 3-22. In view of these fluctuations one is hard-pressed to
explain how farmers can keep up with annual assessments of whether or not fertilizer
use is worthwhile (add to that the uncertainty of climate and actual output prices for

cereals).

 

26Farmers in Ndiapto tended to advance slightly higher prices than farmers in N doga.

99
TABLE 3-22: RECENT OFFICIAL FERTILIZER, PEANUT AND CEREALS PRICES

 

 

 

Fertilizer Price of Pnt./Fert Price of Cer./Fert
Price Percent Peanuts Price Cereals Price
Years (FCFA/kg) Change (FCFA/kg) Ratio (FCFA/kg) Ratio
1976-82 25 42-60 1.7-2.4 35-50 1.4-2.0
1983 50 100% 50 1.00 55 1.10
1984b 90 80% 60 0.67 60 0.67
1985 105 17% 90 0.86 60 0.57
1986c 64 - 39% so 1.41 60 0.94
1987 72 13%. 90 1.25 70 0.97
1988 80 11% 70 0.88 70 0.88
1989 88 10% 70 0.80 70 0.80

 

Sourgg: Kelly (1988b, p.16), with additional calculations and recent

updates from Lg Solgjl. Pre-1984 data are estimates based on various

official sources.

a. Coarse grains (millet, sorghum and maize).

b. Based on the retenue system.

c. Prices in 1986 through 1989 reflect USAID’s declining fertilizer
subsidy of 24, 16 and 8 FCFA/kg.

If one argues the adoption of fertilizer involves a "yes-no” decision, rather than
how much fertilizer to use on a given crop to maximize expected profits, the above
information for fertilizer (i.e., excluding farmers not willing to use improved seeds at 210
FCFA/ kg) can be used to estimate fertilizer adoption rates at alternative prices (Figure
3-4). The gaph suggests most household heads would use fertilizer if it cost 65
FCFA/kg. In comparison, cereals traders sampled believed they could double fertilizer
volumes sold if prices were lowered from 90 to 60 FCFA/kg.

If cash sales of improved inputs are to be encouraged under the APS, fertilizer
should be made available during the cash crop marketing period, when rural liquidity
levels are high (Figure 3-5).27 In 1987 fertilizer was made available only prior to
planting in the research areas (May-June), and the historical experience is that fertilizer

more often than not arrives too late at the farm-level to be utilized prior to seeding (see

 

27Virtually identical results are reported in Crawford et al., (1987).

100

FIGURE 3-4: PREDICTED FERTILIZER ADOPTION RATES AT

FCFA/kg 150
140
130
120
110
100
90
80
7O
60

Sourog: ISRA/MSU FSP Surveys, 1986/7

ALTERNATIVE FERTILIZER PRICES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

North \{outh
\.
\\\
\\\\,
\
\\\ \
\)
0 20 40 60 100%

Cumulative Percent of Farm Households Adopting

Note: Based on the percent of households willing
to use fertilizer at alternative prices; assumes
an output price of 70 FCFA/kg.

101

FIGURE 3-5: NHEN FARMERS HOULD PREFER TO PURCHASE FERTILIZER
AND IMPROVED SEEDS, ACCORDING TO MODE OF PAYMENT

 

Mode Percent Period*

 

 

935') 91% — A
7% - B
1% I C
1% I 0

Credit 6%

30%) IlIIlIIIIlIlIlIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 3
10% C
5%

O

>

 

! ! ! ! !
0 80 120 160 200
Frequency

 

«b 0'

 

*Code for Periods: Peanut/Cotton marketing period (Jan-Mar)
After A, before C (i.e., April)
Beginning of rainy season (May-June)

Indifferent/throughout the year

A
B
C
D

 

Souroo: ISRA/MSU FSP Surveys, 1986/7

Crawford et al., 1987, for an assessment in 1984; this study did not examine the perfor-
mance of SODEFITEX, however). Given the rule of thumb used by farmers discussed
earlier, however, the fertilizer was not "late" if it arrived prior to the emergence of
plants.

3.2.6. Issues Involved in Privatizing Input Distribution

The data suggest that physical access to fertilizer, and credit to obtain it, are as
important as its perceived profitability (confirming findings reported by Kelly, 1988a);
farmers not using fertilizer did not advance a significantly lower yield response expecta-
tion, although they may have been contemplating alternative cereals output prices and/ or
degees of riskiness. Private trader surveys reveal that an important issue is the initially

high price of fertilizer distributed by private traders, which we estimate will exceed the

102
current price of approximately 90 FCFA/ kg by 20-30%, depending on the extent to
which private traders are held responsible for the non-repayment of loans.28

With a higher price of fertilizer, (a) the fertilizer benefit-cost ratio falls below
that calculated officially; (b) farmers will have increasingly superior investment alterna-
tives (Figure 3-3); and (c) they can be expected to not only use less fertilizer per unit of
land but cease to use it entirely (Figure 3-4). As a consequence we may expect the
already unequal distribution of fertilizer to become more unequal, and geogaphically
isolated and agonomically marginal areas not to receive fertilizer at all. The issue for
national policy then is, to what extent should fertilizer use be promoted on marginal land
and remote areas of Senegal?

It does not appear that SODEFTTEX and SONACOS will reduce their activities
in southeastern Senegal in the foreseeable future, making it even more difficult for
private traders to establish themselves. One strategy could involve encouraging private
traders to form a market parallel to that provided by the parastatals, delivering fertilizer
early in the post-harvest period and buying up surpluses of cereals at the same time.
Delivering inputs and buying outputs in one rather than two visits to rural areas would
reduce marketing costs, but would require that farmers leamuand have facilities-40
store fertilizer correctly.29

This raises the broader question, how are the rights of difierent input suppliers

(plant geneticists, breeders, cereals assemblers, wholesalers, retailers and parastatals) to

 

28Nearly one-quarter of the farmers surveyed felt they were not required to repay
fertilizer loans when the crop which had received the fertilizer failed, for example, due to
drought.

29Gaye (op. cit.) alludes to farmers’ concerns about toxic effects of urea on ruminants
and children, a concern seconded by village chiefs in the FSP sample. Furthermore, urea
has a propensity to become compact under humid conditions, making storage more difficult
(see also Crawford et al, 1987).

103

be defined and structured so that conflict (opportunistic behavior)30 is minimized and
the transition to a more productive system completed as quickly as possible? Two
related issues include: (a) to the extent that the adoption of draft equipment is possible
primarily by tying it to a cash crop and an entity that is sufficiently large to absorb debt
losses through transfers from other areas of its operation (such as output marketing or
processing); (b) can traders achieve sufficient scales of operation, without becoming
monopolists or monopsonists, to act effectively as agents of technological change? What
are the knowledge needs of traders, and who is to provide them with that knowledge so
that the learning curve may be compressed? We will return to some of these issues in
the concluding chapter.
3.3. INTERDEPENDENCIES AMONG INPUTS

Earlier in this chapter the organization of rural households and their access to
non-family labor was argued (and seen) to depend strongly on the peanut seed economy.
The purpose of this section is to answer the question, “what difl'erence does the organi-
zation of households make for cereals production performance and the adoption of
mineral fertilizer?" Answers to these questions will reveal some of the consequences of
scaling down the official peanut seed credit progam.

3.3.1. Model Specification

To test the hypothesis that households with additional laborers seed and, irnplicit-
ly, produce more coarse gains than are needed to meet the food requirements of those
laborers, equation [1] was estimated using an instrumental variables (IV) estimator. The
regessors include household structural variables and identifying variables such as the
household head’s age, the value of agicultural draft equipment per active worker, the

quantity of rice seeded and participation with parastatal crop production progams.

 

30Many farmers were not aware of the basic price and interest they were paying for
certain improved inputs received on credit.

104
Since decisions about the two latter variables are unlikely to be made independently of
the quantity of coarse gains seeded, predicted values from two auxiliary regession
equations (using predetermined variables in the system) are used in equation [1]. It is
assumed that decisions related to labor use and household organization are made based
on the amount of peanut seed offered by the household head and other economic consi-
derations such as equipment availability and sharing arrangements. Hence these deci-
sions are independent of, but affect, the quantity of coarse gains seeded. Prices of
coarse gain seed are excluded due to a lack of geogaphic price variation during the

planting season.

[1] CGSEED = a0 + aIVILAB + at2 COND + a3 HIVI + a4 COREM
+ aSCOREF+ aGNFML+ a7HAGE + aSEQMT
A A
+a95TCK+a10RIZS+a11PARA+ ‘1

The following are variable definitions and expected sigrs () for the parameter

estimates oi:

CGSEED - _kg of coarse gain seed used (millet, sorghum and/or maize);

VILAB(+) a vertically integated labor (i.e., that associated with married sons and their
dependents living in the household), expressed in adult-equivalent producers
(AEPs); this variable measures vertical integation (VI);

COND(?) - 1 if the household is horizontally integated (HI), i.e., associated with a
concession, 0 otherwise;

HIVI(?) s the interaction between VTLAB and COND, i.e., HIVI > 0 for households
that are both vertically and horizontally integated; 0 otherwise;

COREM( +) a core labor force-includes the household head’s labor and that of his
immediate male relatives, measured as AEPs;

COREF( +) a the core female labor force of the household, measured as AEPs;

105

NFML( +) a non-family workers and their family members, expressed in AEPs;

HAGE(-) :- the age of the household head—the sign is negative to reflect experience;

EQMT( +) a the value of equipment owned by the household per AEP;

STCK(?) - 1 if food stocks out of the prmding year’s production lasted less than 10
months, zero otherwise; the sign of this coefficient will be negative if insufficient
carry-over of seed reduces quantities seeded and positive if households strive to
not run out of food stocks in the subsequent year;

RIZS(-) - kilogams of rice seeded-«his variable is predicted based on a separate
regession (see note below Table 3-24); to the extent that coarse gains and rice
are substitutes in the household’s cereals budget the sig will be negative.

PARA( +) =- 1 if the household participates in SODEFITEX’S cotton and/ or maize
progam, SODEVA’S maize production progam, and/or SODAGRI’S rice
production progam, and 0 otherwise; this is also the predicted value from a
separate regession on predetermined variables.

The error term 61 is assumed to be independently and identically distributed as N(0,01).
Equation [2] relates the value of NPK and urea fertilizer (FERT) used by

households to the same regessors as in [1] except for the quantity of rice seeded, and

with the additional variables listed below.

[2] FERTsBo-FBlVIIAB+82COND+B3IflVI+B4CORE
+ BSFEML-l- B6NFML+ B7HAGE + BSEQMT+ 89LCON
+ 310 OFFY + 311 FULN + 812 STCK + 813ASCOT
+ 814 SMTAZ + 815 SEVA + £2
The following additional variables are included in the fertilizer equation.
LCON( +) -- measures whether or not the household faces a land constraint—LCON( +)
=- 1 if the household faces a constraint, 0 otherwise (determined from Table 4.3.

in Chapter IV);

106

OFFY( +) a 1 if the household has off-farm income which can serve as a buffer to repay
loans in the case of harvest failure, 0 otherwise;

FULN(-) - 1 if the household is of the traditional livestock herder’s ethnic goup
(Fulani), which may prefer to use manure instead of mineral fertilizer on crops, 0
otherwise;

SCOT(+); (SMAZ(+)); and [SEVA(+)] . 1 if the household is affiliated with SODE-
FITEX’s cotton, (SODEFITEX’S maize) and / or [SODEVA’s maize] production
progam, and 0 otherwise; it is difi'icult to gow cotton and maize without mineral
fertilizer, so these three variables serves as identifying variables. Since there is
also a possibility of a simultaneous equations bias, the instrumental variables
technique was also used on these variables and their predicted values entered
into equation [2].

Since about 50% of the households surveyed did not use mineral fertilizer, the
tobit method is required to estimate [2]. Since this estimator assumes, however, that the
same set of variables affects both the probability and quantity of fertilizer use, two
separate equations are estimated. The first is a standard probit model predicting the
probability (yes or no) of fertilizer. The second equation is estimated on the set of
households using fertilizer, and the error term is assumed to be truncated at zero. The
latter equation pertains to fertilizer quantities used conditional upon use. Both of these
are compared with standard tobit results in Table 3-25.

3.3.2. Results and Discussion

A key question to be answered here is whether or not additional workers allow
the household to produce an additional amount of food that at least equals the con-
sumption requirements of the workers and their non-working dependents. Table 3-23
shows the mean number of workers in the difi'erent categories as well as the dependency

ratios associated with each of the categories. Clearly, additional food requirements are

107

 

 

TABLE 3-23: LABOR CATEGORY SIZE AND EFFICIENCY MEASURES

 

 

 

 

Region
Sample

Category/ Measure North South Total
Core Labor

Mean 3.60 357 3.58

StErr .18 .17 .13
Dependency Ratioa

Mean 1.62 1.63 1.62

StErr .05 .03 .03
Valid N 81 117 198
v1 Labor”

Mean 2.51 3.00 2.79

StErr .35 .30 .23
Dependency Ratioa

Mean 1.42 1.45 1.44

StErr .08 .07 .05
Valid N 23 30 53
Non-Family Labor

Mean 2.30 1.25 2.05

StErr .22 .18 .18
Dependency Ratioa

Mean .67 .96 .74

StErr .06 .12 .06
Valid N 38 12 50

 

i

Sousa: ISRA/MSU FSP Surveys, 1986/7
Note: a. Ratio of adult equivalent consumers to producers.
b. Vertically integated labor force.

108
lower for non-family workers (with a dependency ratio of only 0.74). Nawetaans, for
example, remain in the household for less than 5 months per year.

Table 3-24 presents coefficient estimates for the coarse gain seed equation. With
the exception of female labor, all coefficients estimates for household labor structure
variables are significant at the 10% level or lower. The relatively low R-square value
suggests much of the variation in the quantity of coarse gains seeded remains
”unexplained“. Assuming an average seed-output conversion ratio of 1: 100 for coarse
ga' 31, the results show households with one more adult equivalent producer belonging
to the married son (VI) category gow an additional 332 kg of coarse gains, while an
additional non-family worker adds 364 kilogams. Given per capita (milled) coarse gain
requirements of 200 kgs/annum, which is roughly equivalent to 200 kg of unmilled
coarse gains per adult-equivalent-consumer (i.e., the definition used here),32 these
additional workers allow the household to produce a surplus of food above the needs of
these workers. In the case of married sons the surplus is (332 - 1.44‘200 -) 44
kg/AEC,33 while in the case of non-family labor it is (364 - .74a2oo .) 216 kg/AEC.
Horizontally integated households (HI) produce 800 kg less coarse gains, m
m suggesting that membership in a concession leads to disincentive effects with
respect to food production. On the other hand, households that are both horizontally

and vertically integated produce about two-thirds of a metric ton of coarse gains more

relative to other households (HI‘VI variable).

 

31See Martin, 1988. Estimates for the FSP sample suggest this seed conversion ratio was
attained in the northern, but not in the southern research area.

32The rate of converting unmilled to milled coarse gains is assumed to be 78% (FAQ,
1984). At the same time, there are an average of 0.75 adult equivalent consumers per
person in the households sampled.

33The break-even seed-to-output conversion ratio required for this category is
(ZOO/3.32s) 1:60.

109

TABLE 3-24: COEFFICIENT FSI‘IMATES FOR FACTORS
AFFECTING THE QUANTITY OF COARSE GRAINS SEED-
Dependent Variable= Kg of Coarse Grains Seeded

 

 

 

(Absolute t-statistic)
IV
Independent Variables Estimator

Constant 11.8
(1.49)
VI Labor (VI) 3.32"
(2.08)
Concession (HI) -8.05“
(1.71)
Interaction (HI‘VI) 6.60"
(2.67)

Core Labor: Male 305*
(1.59)

Core Labor: Female .753
(.296)

Non-Family Labor 3.64“
(1.77)

Household Head Age .134
(.942)
Equipment Value/ Worker .720”
(1000 FCFA) (2.00)
1985 Stocks < 10 Months 7.31‘
(1.83)

Parastatal Member@ 7.64
(1.22)

Kg of Rice Seeded@ .0531
(.316)

Number of Households 195
Adjusted R-square 24.1
F-Value (11,183) 6.61“

 

‘(”)-Sigrificant at 10% (5%) or lower.

@: Predicted values of the variables. The set of instru-
ments includes all of the predetermined variables shown
here.

1 10

It is not clear at this point why these results are obtained. Presumably one
explanation is related to different food production risk-sharing arrangements across and
within difl'erent households and mm as alluded to above and as discussed in
Binswanger et al. (1987). For example, household heads in horizontally integated
households may rely on each other and / or the gonoersion head in the case of field-
specific production failures, thus creating a moral hazard34 problem (or social trap).
This does not explain, however, why households that are both vertically and horizontally
integated seed significantly more coarse gains than the other households.

The results in general support the hypothesis stated earlier: household heads
who are able to retain married sons in their households and/ or attract non-family
workers seed sigrificantly larger quantities of coarse gains than those unable to do so,
and the increased output in principle exceeds the consumption requirements of the addi-
tional workers.

Table 3-25 presents results for the equations relating household structure to the
use of fertilizer. Many of the regessors are statistically insignificant, and the probit
equation appears to have difficulty predicting the probability of fertilizer use. There are
two labor variables significantly affecting the mom! of fertilizer use: 1) married sons
(VI labor), suggesting that vertically integated households are more likely to use fertili-
zer (independently of whether or not they gow cotton and /or maize); and 2) HI‘VI
interaction. Households that are both vertically and horizontally integated are signifi-
cantly less likely to use fertilizer. The tobit coefficient estimate suggests households with
one more core female worker are both more likely to use fertilizer and, conditional upon

using fertilizer, use a larger quantity of fertilizer. Households with non-family workers

 

34Layard and Walters (1978, p383) provide the following definition: "There is a
problem of moral hazard whenever the liability of the insurance company is affected by
actions of the insured party about which the insurance company has incomplete information;
the state of nature is thus unobservable by the insurance company." (in the present
discussion the oonooosion head represents the insurance company).

111

TABLE 3-25: MAXIMUM LIKELIHOOD ESTIMATES FOR FACTORS
AFFECTING THE USE OF FERTILIZER

Dependent Variable = FCFA Value of Fertilizer Used
(Absolute t-statistic)

 

 

 

 

Estimator
Independent Variables Tobit Probit Trunc
Constant -13834 -.2582 -51412"
(1.38) (.483) (1.87)
VI Labor (Vertically Integated) 3068" .1743” 1103
(1.93) (1.92) (.390)
Concession (Horizontally Integ.) 54.2 .180 -10165
(.012) (.722) (.964)
Interaction (HI’VI) -3738" -.342"* 267
(1.48) (2.19) (.071)
Core Labor: Male 1605 .103 428
(.865) (.967) (.111)
Core Labor: Female 2978" .141 2233
(1.61) (1.37) (.593)
Non-Family Labor 178 .0628 -1159
(.101) (.627) (.369)
Household Head Age -61.2 -.0126* 732‘”
(.411) (1.58) (2.02)
Equipment Value/Worker 520’ .0137 1.09"
(1000 FCFA) (1.44) (.678) (1.48)
Land Constraint 4764 .161 8038
(.867) (.533) (.768)
Non-Agicultural Income 1817 .307 -9330
(.406) (1.26) (-1.03)
Fulani Ethnic Group -4561 -.306 5066
(1.07) (1.36) (.558)
1985 Stocks < 10 Months -1757 .0443 -10910
(.417) (.197) (1.13)
SODEFITEX Cotton@ 27730’” 1.62‘" 19402
(3.68) (3.20) (1.38)
SODEFITEX Maize@ 1897 " .514 37113‘"
(1.81) (.768) (2.19)
SODEVA Maize@ -4805 .122 -17876
(.262) (.124) (.502)
Number of Households 198 198 94
Log-Likelihood Value 1131 118 1002

Prediction Success: p30 [pa 1]

 

 

*(")[*"]=Significant at 15% (10%) [5%] or lower. @: Predicted values

of the variables. See note for Table 3-24.

71% [57%]

 

112

are not more likely to use fertilizer (the coefficient estimate in the truncated regession
is in fact negative). We may also note, finally, that the married sons and core labor
variables do not significantly afl'ect the quantity of fertilizer used, conditional on using
fertilizer, for this sample.

Older household heads and members of the Fulani ethnic group are less likely to
use fertilizer, although households with older heads use a sigiificantly
larger quantity of fertilizer once they have decided to use it. In the specification shown,
the sign on the non-agricultural income variable is positive in the probit model, but
negative in the truncated version. As hypothesized earlier, a household’s ability to draw
on non-agricultural revenues to repay production credits in the event of crop failure may
make it more willing to invest in fertilizer, while using less-once it has decided to use
fertilizer-relative to households without off-farm income that use mineral fertilizer. In
summary, the coefficient estimates in Table 3-25 suggest relative input/ output prices are
not the only factors affecting the use of mineral fertilizer by farm households in Senegal
(see also Kelly, 1988a).
3.4. SUMMARY AND CONCLUSION

There is a considerable diversity of resource ownership among farm households
in southeastern Senegal. While relaxing (ash constraints is important-especially in the
pre-planting period as envisioned under the APS-not all households will necessarily be
best served with fertilizer and improved coarse gain seed inputs. Access to ash crop
seed (peanuts) appears to be important in attracting workers and/or retaining them in
the household, which leads to a larger quantity of coarse gains planted. The ability to
retain married sons also appears to affect the willingness of households to use fertilizer
(controlling for the gowing of cotton and maize, which require fertilizer), but no such
relationship was found in the case of non-family labor. Access to and afl'iliation with

parastatals is of some importance in "explaining" equipment ownership and use, which is

113
also interdependent with the size of the household’s labor force and, to some extent, its
structural organization.

While more work is needed on the determinants of household structure and the
relationship between household structure and performance, the results presented here
suggest that viewing households as a coalition of individuals with different interests may
lead to a more solid microeconomic foundation for understanding the organization of
Sahelian agiculture, where markets commonly do not conform to neoclassieal assump-
tions and intra-household trade can become a substitute for conventional market ex-
change. F or more immediate purposes, the results suggest that, under current conditions
of coarse gain market uncertainty, the "cash-versus-food" crop issue poses less of a
dilemma than previously imagined. This results from the linkage between cash crop
seeds and additional laborers attracted with the seeds, who in turn also work on the
collective cereals field of the household. Complete government withdrawal from peanut
seed distribution, without the development of an effective private sector replacement dis-
tribution system that can offer seed on credit, may well lead to lower rather than higher
national food production and self-sufficiency levels, by inducing smaller labor forces in

rural households.

CHAPTER IV
HOUSEHOLD PRODUCTION BEHAVIOR AND POSSIBILITIES

This chapter describes the crop and non-crop production activities of households
sampled; their anticipated responses to the NAP; prospects for expanded maize produc-
tion; and recent changes in the crop mix gown by households. In this chapter a set of
feasible production strategies is identified and an evaluation is made of the "trade-off‘
between two key strategies, namely food crop and cash crop production. This is done by
means of an estimated cost function, which is used to calculate economies of scope and
scale in producing the two types of crops. The next chapter examines the relationship of
the strategies identified here to the food security performance of individual households.
The final chapter draws implications of the results presented in this chapter for national
crop research priorities and strategies.
4.1. CROP PRODUCTION GOALS, STRATEGIES AND RECENT DYNAMICS

Section 4.1. examines crop mixes and production levels of households in south-
eastern Senegal (using official and survey data), factors correlated with different crop
production levels, and primary constraints to expanding areas cultivated advanced by
household heads. Given the importance of maize in Senegal’s new food strategy, section
4.1.4. reports maize production problems reported by farmers. The section concludes by
presenting crop mix changes in 1987 relative to 1986 along with reasons given by farmers
for these changes, and evidence related to their crop production decision-making.

4.1.1. Crop Mixes and Production Levels

Most households (98%) gew millet, sorghum and/ or maize in 1986, while over
two-thirds of the households in the south also gew low-land rice (Figure 4-1). Figure 4-
2 suggests that in comparison to all of Senegal, farmers in the southern part of the study

zone allocate their land to a geater variety of crops. To a large extent this reflects

114

115

 

 

PERCENT OF HOUSEHOLDS 0mm

 

FIGURE 4-1: CRORs GROWN IN THE FSP RESEARCH AREAS
BY NORTHERN AND SOUTHERN REGION. 1986
[PERCENT OF' HOUSEHOLDS GROWING EACH CROP]
LEGEND
Noam

- scum

MILLET' HM MAIZE RICE COTTON
SORGHUM FD MAIZE PEANUTS
WPE OF CROP

Source: ISRA/MSU FSP Suwoyo. 1956/7

 

 

 

 

 

 

PERCENT OF TOTAL HECTARAGE

FIGURE 4-2: CROP MIXES IN SENEGAL AND THE
FOOD SECURITY PROJECT RESEARCH AREAS

 

 

LEGEND
@ Rico
E Maize
sorghum
Millet
- Cotton
Peanut:

 
  

 

  

7
/
Z

SENEGAL theul MOkO Kounk DObO/Dcln
\_"Northern“_/ \_"Southern"_/
ARRONDISSEMENT/STUDY AREA
SOURCE: GOVERNMENT OF SENEGAL. MOP/DA. 1966

 

 

116
increased production opportunities owing to higher rainfall; the cultivation of sorghum,
for example, is possible only in the relatively rain-abundant southeastern areas of the
country (the role of tastes is examined in Chapter V, Section 5.1.4.).

Table 4-1 shows crop production levels in the two survey regions. Valued at
official 1987 prices,1 the goss market value of total crops produced per AEP was 60%
higher in the north (106,000 FCFA/AEP) than in the south (66,500 FCFA/AEP). One-
quarter of the households produced 47% of all coarse gains in the north, and 60% of all
coarse gains in the south. This appears to reflect the somewhat more even distribution
of equipment in the north (see Chapter III).

In a year of adequate rainfall 66% of the households in the south failed to
produce sufficient quantities of millet-sorghum-maize (estimated at 200 kg per adult-
equivalent consumer, AECZ) to carry them through until the subsequent harvest (Figure
4-3). Adding rice production reduces this percentage only marginally (to 60%). In
comparison, only 20% of the northern households produced less than 200 kgs of cereals
per AEC. Information obtained for the 1987 crop year suggests the food situation
observed in 1986 was not unique: sixty-four percent of the southern household heads
reported they had not seeded enough cereals (millet, sorghum and maize) in 1987 to
satisfy household consumption needs. Thirty-six percent indicated they had seeded
enough (30%) or more than enough (6%). In comparison, only 11% of the northern
household heads reported they had not seeded enough, while 34% indicated they had
seeded enough to meet household needs and 54% responded they had seeded enough to

produce a surplus of cereals.

 

1These are cereals:- 70 FCFA/kg, peanuts =- 90 FCFA/kg, cotton =- 100 FCFA/kg and rice
8 150 FCFA/kg. The latter is an estimate representing a mix of official and unofficial rice
and paddy rice prices.

28cc also the discussion in Chapter III on the use of consumption requirement
standards (Section 3.3).

117

TABLE 4-1: NEAN ANNUAL HOUSEHOLD 059? PRODUCTION,
BY STUDY REGION, 1986

(Standard Error)

 

 

 

 

Production_iK9sl Total
Study Coarse d ’000 FCFA
Region Grains Ricec Peanuts Cotton Value
North
Total 2978 1 3175 169 514.6
b (230) (1) (455) (80) (54)
Per AEP 676 0 605 56 106.1
(55) (0) (75) (31) (9)
South
Total 1317 213 896 536 255.3
(146) (27) (126) (101) (25)
Per AEP 326 52 248 139 66.4
(32) (5) (34) (25) (5)

 

Sggngg: ISRA/MSU FSP Surveys, 1986/7

Notes:

a. Minor crops such as fonio, cowpeas and cassava are excluded.
According to official 1986 estimates, these crops occupy less than
1% of the land in the research areas.

0'

. See Table 3-3 for definitions of AEPs.

c. Paddy rice is converted assuming a 65% transformation rate.
d. In Thioubouk (Kounkane, Velingara) 8 fully equipped households
produced an average of 2,950 kg of cotton each (equivalent to 640

kg/AEP).

e. Prices used are: Coarse grains - 70 F/kg; rice - 150 F/kg; peanuts
- 90 F/kg; cotton-100 F/kg.

118

 

FIGURE 4-3: CUMULATIVE FREQUENCY DISTRIBUTION FOR
PER AEC CEREALS PRODUCTION. BY STUDY AREA. 1986
(In Consumable Product Equivalents)

 

3

 

I
/

 

North

§

 

 

 

 

§§§§§
l

 

 

5

K65 0F CEREALS PER KNIT CONSUMER EOINALENT

 

 

 

A A A A A A A A A A A A A A FA
r r vvvvvv
v v v v w v v v v v v v v v v v v v v v v v v 1 vvvvvvv

O

CUMULATIVE PERCENT OF FARMS
Sourco: ISRA/MSU FSP Sumyl. 1986/7

 

 

 

Virtually all household heads believe it is preferable to gow enough cereals to
meet home consumption needs rather than buying staples, mainly because cereals
markets are perceived as unreliable during years of food production shortfalls (see also
Chapter VI). Over 90% of the household heads reported cereals consumption needs
were the first factor taken into consideration in determining the area seeded to cereals
(see Goetz, July 1988 and the discussion in Section 5.1.5.). This raises the question why
many households nevertheless failed to produce sufficient cereals in two consecutive
years to meet estimated household food needs.

The following hypotheses are offered at this point to explain why per adult
consumer equivalent cereals production is low in the southern areas. (1) The lack of
equipment, combined with relatively abundant rains leading to rapid weed gowth, entails

a severe peak labor demand which reduces the total area cultivated per worker.3

 

3 Herbicides are not widely used (see Chapter III).

119
(2) The lack of equipment tends to be interdependent with a relatively low supply of
labor at the farm level: a minimum number of active workers is required to profitably
use a given set of agicultural implements, but in households without equipment young
sons are more likely to migate and / or seek off-farm sources of income. (3) Pursuing
off-farm activities (including seasonal labor migation, petty trading and raising of
livestock) is on average more profitable than gowing cereals. (4) Cotton and peanut
crops are preferred because they have secure market outlets at known prices. These
crops-4n contrast to cerealsualso tend to be less prone to attacks by wildlife.

Given insigtificant differences in average crop yields between the two areas
(according to official estimates), Table 4-2 suggests that household crop production per
AEP is related to the level of equipment use, although the value of the output does
differ significantly between non-users and borrowers in the south. Differences between

borrowers and fully equipped households are less pronounced in the north, presumably

TABLE 4-2: ANNUAL CROP PRODUCTION PER AEP AND EQUIPMENT USEa
(Standard Errors)

 

 

 

 

Study Region Productign_ngr_AEE_LKg§) Total
and EquiB- Coarse d ’000 FCFA

ment Use Grains Ricec Peanuts Cotton Value
North

Borrower 712 0 359 148 84.4

(182) (150) (137) (21)

Fully Eq 661 0 665 32 110.3

, (55 (86) (12) (10)
South

Non-User 196 64 309 50 54.8

(44) (9) (105) I31) (12)

Borrower 261 46 193 103 53.8

(26) (8) (35) (27) (5)

Fully Eq 551 53 293 273 98.6

(83) (15) (62) (65) (13)

 

Source: ISRA/MSU FSP Surveys, 1986/7
a. See Table 3-3 for definition of AEPs;
b. Table 3-5 for equipment definitions.

120
due to the geater availability of equipment even for borrowers (note the large standard
error associated with that category, however).

In both research areas households (a) of the Fulani ethnic goup; (b) not facing a
land constraint; and (c) whose 1985 cereals production had lasted nine months or longer
produced more coarse gains per active worker (Table 4—3). In the north (where land is
becoming more of a constraint relative to the south) it appears that farmers shift into
(possibly intensified) production of higher-valued peanuts rather than coarse gains,
when land becomes a constraint and once food needs have been met (see also the dis-
cussion in 4.1.5.).

In the north the following households produced more coarse gains per worker:
households headed by younger individuals; those that were not integated; those using
only family labor; and those not participating with SODEFITEX. In contrast, southern
households that produced more coarse gains per worker were: those which were inte-
gated in some sense; those owning cattle; those who participated with SODEFITEX
and / or SONACOS and did not pursue off-farm activities. Consequently, there are
considerable diflerences between the northern and southern areas, as will be further
confirmed in the following section.

A further analysis, useful also for issues examined in the following chapter, is
carried out by classifying households into three 'production-security' categories according
to the amount of all cereals (milled coarse gain equivalents4 and rice) produced per
adult equivalent consumer (AEP). The categories selected were:

1. Low producers: 0.160 kg per AEP, who are considered to be food

deficit;

 

4A 78% conversion ratio from unmilled to milled coarse gains is assumed (FAO, 1984).

121
TABLE 4-3: VARIABLES ASSOCIATED NITN PER NORKER PRODUCTION

 

 

Coarse Grains (Kgs) Rice (Kgs)
North South South
Variables Mean StErr Mean StErr Mean StErr
Age of Household
Head: 20-39 yrs 744 107 355 71 62 15
40-54 yrs 629 81 270 38 30 6
55-90 yrs 570 65 351 61 59 10
Nmbr of Hives: O 873 178 105 69 46 35
1 725 90 358 46 37 7
2 513 48 315 56 63 12
3 723 216 209 87 78 25
Ethnic-Fulani 794 107 363 39 48 7
Ethnic=0ther 580 52 162 29 51 11
Single Family 729 78 289 34 51 9
Vertical Integ. 549 68 394 112 36 13
Horizontal Int. 642 117 364 99 52 12
HI*VI 437 42 339 106 46 9
Family Lab. only 713 66 323 36 47 6
Use Non-Fm. Lab. 564 74 315 70 64 22
No land constr. 690 69 323 33 49 6
Land constraint 563 61 261 31
’85 stck < 9 mth 534 86 286 42 47 8
GE 9 mths 689 60 380 51 52 10
Cattle Owner 697 80 466 58 54 11
Non-owner 639 88 205 26 49 8
w/ SODEFITEX 531 62 387 61 33 9
not w/SODEFIT. 694 65 281 36 58 8
w/ SODEVA 592 82 . . . .
not w SODEVA 665 61 322 32 49 6
w/ SONACOS 631 72 366 40 57 9
not w/ SONACOS 661 70 273 52 40 8
Off-farm Income 633 64 291 33 57 8
None 668 82 395 74 32 8
Borrowed Cash 634 75 334 72 54 12
No borrowing 657 68 319 37 47 7

 

122
TABLE 4-3: Continued

 

 

Peanuts (Kgs) Cotton (Kgs)
North South North South

Variable Mean StErr Mean StErr Mean StErr Mean StErr
Age 20-39 yrs 601 137 203 45 89 77 195 57
of 40-54 yrs 830 167 247 57 44 28 97 32
Head 55-90 yrs 431 83 260 54 29 14 123 44
Nbr of wives: O 344 265 124 113 101 75

1 607 124 256 42 81 58 165 37

2 580 86 221 53 14 7 121 42

3 1220 461 257 98 43 43 47 47
Fulani 472 107 227 35 26 14 164 31
Non-fulani 654 96 276 65 68 41 34 24
Single Family 712 133 255 37 48 21 105 29
Vertically Int. 558 164 198 113 6 6 190 77
Horizontal Int. 459 101 282 82 109 89 210 78
HI*VI 593 195 103 28 7 7 88 38
Fam. Lab. Only 541 96 225 32 79 48 138 28
Use Non-Fam Lb. 661 115 337 102 22 16 121 63
Non Land Const. 466 81 239 31 69 39 137 26
Land Constraint 857 135 96 20 12
’85 Stock<9mths 311 65 205 36 110 85 127 32
GE 9 months 713 96 293 54 29 12 151 42
Cattle Owner 789 155 235 41 1 l 215 47
Non-Owner 513 115 223 45 111 80 69 23
w/SODEFITEX 475 88 99 18 167 80 336 50
Not w/ $00. 652 100 321 44 16 8
w/SODEVA 532 135 . . 28 16 . .
Not w/ SODEVA 614 88 238 30 61 35 136 25
w/SONACOS 640 102 296 43 84 50 160 35
Not w/ SONACOS 547 108 173 41 19 9 109 37
Off-Farm Inc. 581 97 225 34 7O 41 82 20
None 617 112 267 63 21 11 258 64
Borrowed Cash 578 115 333 85 69 54 56 30
No Borrowing 609 96 208 29 38 17 161 32

 

Sggngg: ISRA/MSU FSP Surveys, 1986/7

123

2. Medium producers: 161-300 kg per AEP, who are considered food

production secure; and

3. High producers: 301+ kg per AEP, who are considered food

production surplus.
In the sample, 32% of the households are estimated to be food-production insecure,
while 31% (mostly in the north) produced enough cereals to last them 18 months or
longer (Table 4-4). Food production-insecure households in the south on average
produced proportionally more rice than the food production secure and surplus house-
holds, possibly reflecting different labor constraints. Households in the "low“ category
produced less cash crops per worker, and they had fewer workers than those in the
medium category and a slightly higher ratio of consumers per active worker (1.64 vs. 1.58
and 1.50 in the ”medium" and ”high" categories). The significant and sizeable differen-
ces in equipment owned per active worker across the three strata are noteworthy.
Nevertheless, 22% of the food production-insecure households were also fully equipped,
while one~quarter of the households in the "high" category were able to produce a
surplus without being fully equipped. Some of the production-insecure households also
used non-family labor, and they were not significantly more or less likely to be inte-
gated (either vertically or horizontally). At the same time, 27% of the integated
households were in the 'low" category, while 29% were in the ”high" category.

Households in the food production surplus category were more likely to own
cattle (relative to households in the low category), and most had also produced at least
enough food to last them 9 months or longer in the preceding year. A fairly large
proportion of households in the low category also were working with SODEFITEX. This
may partially reflect the fact that many of the households in the low category are in the

south, and it is unclear whether they work with the parastatal because they produce

124

TABLE 4-4: PRODUCTION VARIABLES ASSOCIATED WITH NOUSEHOLDS'
FOOD PRODUCTION SUFFICIENCY STATUS

 

 

Kilograms of all Cerealsa Total
r
Low Medium High
Variable 0-160 161-300 301+
Households: No. 45 53 44 142
Percent 32% 37%. 31% 100%
All Cereals (kgs)
Produced/Horker 157 366 718 409
(12) (15) (48) (25)
Rice as a % of
Cereals (South) 24 18 13 20
(5) (2) (4) (2)
FCFA Cash Crops
Produced/Horker 23929 41385 70064 44739
(3275) (5459) (8839) (3868)
Number of Horkers
in AEPs 4.1 5.0 4.3 4.5
(.4) (.4) (.4) (.2)
FCFA Equipment
Owned per Horker 1036 2787 4571 2785
(229) (367) (477) (243)
Percent Fully
Equipped 22 51 75 49
(5) (7) (7) (4)
Percent Using Non-
Family Labor 12 24 35 24
(5) (6) (7) (4)
Percent Integrated
[H1 or VI or HIVI] 4O 58 43 48
(8) (7) (8) (4)
Percent Owning
Cattle 30 54 58 47
(7) (7) (8) (4)
1985 Food Prodn.
lasted 9 mths + 33 48 77 52
(7) (7) (5) (4)
Percent HBrking w/
SODEFITEX 42 26 36 35
(7) (6) (7) (4)
Percent Norking w/
SODEVA (North only) 33 28 19 24
(21) (9) (6) (5)

 

Sggggg: ISRA/MSU FSP Surveys, 1986/7

a. All coarse grain figures are converted into milled equivalents assming a 78%
transformation rate. b. Cotton and/or maize production programs.

125
insufficient amounts of food, or whether they produce insufficient amounts of food
because they work with the parastatal.
4JJL IfintehMMPCbnsUHdnmstolbaensflhnmknl

This section reports interviewees’ responses to shed light on the following ques-
tions: which factors reportedly prevented household heads from expanding areas cultiva-
ted in 1986; to what extent are they influenced by cereals price policies; and how would
they react to a reduction in peanut prices?

Household heads in the north of the study area responded that a lack of labor,
peanut seed and land were principal factors preventing crop extensification in 1986
(Table 4-5). In the south a lack of equipment and labor were cited as the main reasons,
even by respondents in "fully equipped” households. While some households invested in
labor-saving draft equipment prior to the 1987 season, this was not sufficiently wide-
spread to have made a substantial difference for agricultural production in 1987 (see
GH, March 1988).

TABLE 4-5: PRIMARY CONSTRAINTS T0 EXPANDING AREA
CULTIVATEO IN 1986
Percent of Households Responding

 

 

 

Constraint
Region Land Labor Equipment Seeda Other
North 35b 25 5 35
South 11 3O 4O 15 4

 

Sggngg: ISRA/MSU FSP Surveys, 1986/7

a. This is peanut seed in the north, cereals seed
in the south.

b. In most of these cases the land is perceived as
being "too remote to be worth cultivating".

126

The results therefore suggest that credit for purchasing inputs, and not the
unprofitability of production opportunities alone, was a primary constraint to extensifica-
tion (note that the question asked which general factor prevented expansion, i.e.,
allowance was made for price or profitability factors). This does not mean, however,
that simply providing more cash to farmers would allow them to expand production,
since it is not clear that input supplies (labor, equipment and / or seed) would readfly be
forthcoming in any given area. The data in Table 4-5 also suggest that most farmers
would not be forced to intensify agricultural production (increase the amount of inputs
used per unit of land) because they lack suitable land.

As discussed in Chapter III, household heads were also asked which crop they
would have gown more of in 1987 if they had been able to expand production. One-
half of the respondents in the north answered they would have gown more peanuts,
while most (83 %) southern respondents would have gown more millet, sorghum and / or
field maize. This indicates the more food-secure northern farmers view peanut produc-
tion as more profitable than cereals production at current relative input and output
prices (see also Kelly, 1988a and GH, May 1987), while southern household heads would
attempt to make up the food production deficit rather than commit additional resources
to cash crop production.

Most northern household heads responded they would not change their cereals
areas seeded if the floor price were reduced or eliminated, while one-quarter responded
they would not change the area seeded if the price were increased (see also the more
detailed results in Section 5.1.6.[i]; observations for the southern areas have to be
interpreted with caution since farmers have no experience with effective coarse gain
floor prices; evidently, some of the northern respondents may have been considering a
change in the use of inputs per unit of land). The responses suggest it is difficult for

some household headsuas buyers or sellers—to conceive of cereals as a "true“ cash crop.

127

Three plausible explanations are (a) the historical role of peanuts and cotton (with
assured input delivery and output markets), and possibly livestock, as a source of cash
revenue, combined with the only recently ganted right to trade cereals freely and at a
supported price; (b) relative input and output prices are such that it is more profitable
to gow peanuts, even though cereals prices have recently been rising more rapidly than
peanut prices; (c) the "ghost of past famines" (as recent as 1984/5) which, combined with
the unreliability of markets as a source of cereals, makes selling of staple cereals
unthinkable for some household heads.5

If (a) above is the principal explanation, household heads are likely to adjust
their perceptions of cereals as a cash crop over time. In fact, sizeable sales of sorghum
in 1987 in Maka (Figure 5-3 in Chapter V) and sales of all cereals in the Medina Yoro
Foulah area (Kolda) in 1987 illustrate this adjustment may already be underway. As an
indirect and partial test of explanation (b), farmers were asked how they would react if
official peanut prices were reduced from 90 to 70 FCFA/kg (this policy was actually
implemented in 1988). One-half (38/ 78) of the northern household heads reported they
would seed more peanuts, while 28% would not change the area seeded to peanuts. In
the south most farmers responded they would seed less (39% = 42/108) or the same
(41%) amount of peanuts.

In the north most household heads (67%) responding they would seed more
peanuts also indicated they would not change the area cultivated to other crops. They

may have excess productive capacity and would gow more peanuts but not more cereals

 

5This may be related to a cultural norm which stipulates that if a household sold coarse
gains one year, and then had insufficient food supplies the next (for whatever reason), it
was the household that got itself into trouble by selling food and therefore did not deserve
any assistance from other households.

128
to meet target cash needs.6 Those who would seed less peanuts indicated they would
switch mainly to millet and/or sorghum (55%) and maize (35%). In the south some
household heads would reduce millet-sorghum (30%) and maize areas (23%), if they
were to increase the size of their peanut fields. Most (89%) of those responding they
would reduce their peanut areas answered they would instead gow cereals.

Taken together, the results imply that most farmers in southeastern Senegal face
resource and / or credit constraints, and that simply raising cereals prices is unlikely to
have a significant impact on output. In fact, we will see in Chapter V that households
relying on cereals markets to purchase food already face higher cereals prices than
households selling cereals, so that they in the short-run may be adversely affected by a
higher cereals floor price.

To conclude this section, examining household heads’ opinions of-oand stated
responses to-current and future agricultural policies in Senegal is a useful starting point
for informing political decision-makers about the anticipated consequences of those
policies in different regions of the country. Additional information on opinions is
examined in Chapter V. The following section presents results pertaining specifically to
the APS.

4.1.3. Considerations of Cereals Output Response under the APS

On average, household heads indicated they would be willing to replace 67% of
their traditional coarse gain seeds with the improved seeds (range =- 10% to 100%;

proportions are slightly higher for maize than for millet and sorghum).7 This average is

 

6 These farmers may also plan to purchase more peanut seeds at the lower price, still
finding peanuts more profitable than cereals at the lower (but nevertheless gertam' ) price.
Given the low output per seed ratio of peanuts, seed is an important cost of producing
peanuts (see Section 3.1.4. and Gaye, 1986). More peanut seed would also allow them to
attract more workers.

7Results include all farmers, assuming those not willing to pay official prices get the
inputs at their proposed prices.

129
high, given the importance of assuring cereals self-sufl'iciency and, from the farmer’s
point of view, the untested nature of the seeds.

After presenting respondents with per hectare seed and fertilizer application
rates proposed under the APS (e.g., 4 kg of millet seed for every 150 kg of NPK and
50 kg of urea per hectare, and the expected output achievable with these inputs), they
were asked how many hectares they would cultivate to improved cereals (Table 4-6). At

proposed application rates these areas correspond to an average of 730 kg of fertilizer

TABLE 4-6: HECTARES III'IICH FARMERS IIOULD REPORTEDLY CULTIVSI' E
TO IMPROVED MILLET/SORCI'IUM AND MAIZE; ASSUMES CASH SALES

 

 

 

 

Millet/Sorghum Maize
Study
Region Mean Range N Mean Range N Totals
North 1.9 .5-5.0 82 1.4 .3-4.0 83 3.3
South 1.2 .3-5.5 120 0.9 .3-6.5 122 2.1
Total 1.5 .5-5.5 202 1.1 .3-6.5 205 2.6

 

Source: ISRA/MSU FSP Surveys, 1986/7
a. Responses for credit sales: millet/sorghum N-2.3, S-l.4;
maize N-l.6, 5-1.

per northern and 465 kg per southern household, considerably above historical levels
received by households.8 For most farmers-particularly in the southern areas-the use
of improved seeds and fertilizer would entail an increase in the total area cultivated to
cereals (Table 4-7). It is unclear to what extent this increase would reflect cultivation of
new land or a reduction in land allocated to peanuts and cotton.

Table 4-7 confirms the earlier finding that the availability of cereals seed is one

important factor determining hectares gown to cereals in the southern area, especially

 

88cc, e.g., Crawford et al. (1987), and other FSP reports on input use. The calculations
assume actual application rates of 250 kg/ ha on maize and 200 kg/ ha on millet.

130

TABLE 4-7: HOH FARMERS HOULD CHANGE TOTAL AREA SEEDED TO
CEREALS.CONCURRENT HITH RECEIPT OF IMPROVED SEEDS UNDER THE APS
Number and Percent of Households Sampled

 

 

 

Area Study Region

Change North South Total
Increase 54 62% 110 89% 164 78%
Decrease 3 3% 3 1%
No Change 30 34% 14 11% 44 21%
Total 87 100% 124 100% 211 100%

 

Source: ISRA/MSU FSP Surveys, 1986/7

in Sandio and Diega, where 98 % and 93 % of household heads said they would increase
areas cultivated to cereals if they obtained cereals seed under the APS.9 Consequently,
distribution of the improved inputs may, depending on the region, lead to both a yield
and an area emansion effect for cereals.

From the above information approximate average coarse gains yields and
hectares cultivated by households in 1987 can be calculated. The 3.3 and 2.1 hectares
cultivated using improved seeds (Table 4-6) are in fact the estimated areas gown in
1987, since they represent 67 % of all cereals seeded in the northern and southern areas;
assuming identical seeding rates for improved and traditional seeds, the total area culti-
vated per household would therefore be 4.9 and 3.1 hectares, 33% of which represents
an increase attributable to the APS. Assuming no change in yields and average areas
cultivated to cereals in 1987 relative to 1986,10 this gives a yield estimate of 905 kg/ha

for northern and 630 kg/ ha for southern households.

 

91p part this may also reflect the higher expected profitability of the cereals, however,
rather than only a relaxed input constraint.

10Production data for 1987 were not collected.

131

The yield estimate for the northern area is close to official 1986 estimates while
the southern estimate is below these estimates. In part the discrepancy may be
explained by a) yields and areas cultivated did change between 1986 and 1987, b)
farmers have a more difficult time visualizing crop area changes in the south and/ or c)
the lower southern yield reflects higher crop damages by predators and pests and/ or
increased weeding problems.

Not all farmers are likely to adopt the entire technological package. In
particular, an estimated 14% of the household heads will adopt only fertilizer or
improved seeds, but not both at the same time, regardless of whether sales are on a cash
or credit basis.

4.1.4. Prospects for Expanded Maize Production

Since maize is especially targeted under the APS, it is worth examining reasons
given by farmers for gowing maize as well as factors currently affecting maize produc-
tion, to see if maize supplies are likely to increase in response to price incentives. At
the national level, according to official statistics, the area and production of maize has
increased since 1980 (Figure 4-4). As discussed briefly in Chapter I, a shift-share
analysis for the period 1961-85 reveals that the share of the Peanut Basin in total
national maize production is increasing, while the shares of Senegal Oriental and the
Casamance are declining, despite their higher rainfall (Goetz with Dieng, March 1987,
p.11). To some extent this may reflect a response to the recent drought since the newer,
ear -maturing maize varieties appear to serve as a food security crop during the hungy
season.

In 1987, over 95% of the households sampled gew maize in the traditional field
around the compound or as field maize. Due primarily to land pressure, there is a
tendency to gow maize further away from the home in the northern areas. Farmers

generally believed they could have increased yields in their home maize field (56%), but

132

 

FIGURE 4-4: MAIZE PRODUCTION AND AREA CULTIVATED:
SENEGAL: 1961 - 1985

Prdn (‘000 kg.)

 

...... - Aroa (hoover-O)

III 1000 Kgs and Heciotes

 

o
1961 1966 1971 1976 1951 1986
Year

 

 

SOURCE: USA": Date: Boo.

 

not expanded the area cultivated (69%). Most farmers either did not plan to use
fertilizer on their field maize (52%), or had not acquired it at the time of the survey
(July, 1987). Nevertheless, fertilizer was identified as an important input into maize
production. Maize gown around the compound usually benefits from animal manure
and "night soil” and therefore does not require commercial fertilizer.

Most farmers gow field maize as a food crop, destined for own-consumption and
not for cash sales (Table 4-8).11 Farmers not gowing field maize cited, as main
reasons, a lack of fertilizer in the northern, and problems with natural predators in the
southern areas. Potential output marketing constraints and ways of reducing these to

stimulate the production of maize are explored in subsequent chapters.

 

11Household preferences among different cereals, and storage and transformation
constraints, are discussed in Chapter V.

133

TABLE 4-8: REASONS FOR AND AGAINST CROHIMB FIELD MAIZE
Percent of Households Responding

 

Study Region

 

 

 

Responses North South Total
Reason for Growing

Serves as our Food ....... 36% 73% 56%
Gives Good Yields ........ 7% 24% 16%
Home Maize is not

Sufficient ............... 17% 2% 9%
Lack Land for Home Maize. 14% 2% 8%
Sells Easily ............. 17%» 0% 8%
Other Reason ............. 10% ‘ 0% 4%
Reason for not Growing

Monkey/Harthog problem... 0% 58% 36%
Lack Fertilizer .......... 50% 0% 19%
Lack Labor ............... 19% 23% 21%
Lack Suitable land ....... 12%) 6% 8%
Lack of Market Outlet.... 5% 6% 5%
Other Reason ............. 14%) 7% 10%

 

Source: ISRA/MSU Food Security Project Survey, 1987
Note: The question allowed only for the one most
important reason for growing or not growing field maize.
This does not mean, for example, that soil fertility is
not a concern in the southern research areas (see also
Table 4-9).

Attacks by natural predators, high fertility needs and high labor demands were
listed as the main general problems with producing field maize by those gowing the crop
(Appendix A-4). High moisture requirements of existing varieties were cited as a
concern in the northern but not in the southern areas. Labor and tools for land prepara-
tion along with a lack of fertilizer constituted principal constraints to expanding the area
cultivated to maize in 1987 (Table 4-9).

One goal of the NPA is to induce farmers to substitute maize for millet and
sorghum (allegedly less fertilizer-responsive). Those farmers not willing to make the

reduction indicated that millet and sorghum were their principal staple, maize production

134

TABLE 4-9: FACTORS PREVENTING EXPANSION OF FIELD MAIZE CROPS
Percent of Households Responding

 

Study Region

 

 

 

 

Factors North South Total
First Reason

Labor/Field Hork ........... 32% 43% 38%
Lack Fertilizer/Manure ..... 38% 6% 22%
Lack Equipment/Animals ..... 12%. 25% 19%
Lack Suitable Land ......... 12% 8% 10%
Lack of maize seeds ........ 2% 8% 5%
Labor/Crop protection ...... 0% 6% 3%
Lack Land in general ....... 2% 2% 2%
Rainfall Insufficient ...... 2% 2% 2%
Second Reason

Lack Equipment/Animals ..... 46%} 22% 32%
Labor/Crop protection ...... 3% 39% 23%
Labor/Field Nork ........... 18% 18% 18%
Lack Fertilizer/Manure ..... 15% 8% 11%
Lack of maize seeds ........ 0% 10% 6%
Rainfall Insufficient ...... 8% 2% 4%
Lack Suitable Land ......... 3% 0% 1%
Lack Land in general ....... 3%. 0% 1%
Other Reason ............... 5% 2% 3%
Third Reason

Rainfall Insufficient ...... 19%) 29% 26%
Labor/Field Hork ........... 19% 1N% 17%
Lack Fertilizer/Manure...;. 24%» 17% 19%
Lack Equipment/Animals ..... 19% 8% 12%
Lack of maize seeds ........ 5% 13% 10%
Labor/Crop protection ...... 0% 13% 9%
Lack Suitable Land ......... 5% 2% 3%
Lack Land in general ....... 0% 2% 1%
Other Reason ............... 10% 0% 3%

 

Source: ISRA/MSU Food Security Project Survey, 1987

135
is "too risky", and that the land currently cultivated to millet was not suitable for maize
(Table 4-10). Farmers willing to substitute maize for millet and sorghum, would do so if
they had more fertilizer (especially in the North), if the price of maize rose relative to
that of millet and sorghum, and if they had more maize seeds (Table 4-10).

TABLE 4-10: NILLINGNESS T0 SUBSTITUTE MILLET/SORGHUM FOR MAIZE
Percent of Households Responding

 

Study Region

 

Willingness North South Total

 

A. Not willing to
Substitute because...
Mil/Sorghum is our essential

 

staple .................. 65% 36%) 44%
Maize is too risky ......... 8% 40% 31%
Land not suitable .......... 15% 11%. 13%
Insufficient

labor/equi pment ......... 0% 10% 7%
Other ...................... 12% 3% 5%
B. Nilling to substitute

f. O 0
Had (more) Fertilizer ...... 60% 35% 51%
Relative price rises ....... 33% 6% 22%
Had more maize seeds ....... 4% 35% 16%
Less predators ............. 0% 12% 4%
Other ...................... 4% 12% 7%

 

Source: ISRA/MSU Food Security Project Survey, 1987

In summary, the following conclusions can be drawn regarding the prospect of
expanded maize production in southeastern Senegal (1) The potential for expanding
maize production in traditional fields around the compound is limited. Substantial gains
are likely to be possible only by promoting field maize, but this has to be accompanied
by a sound technological package to help farmers overcome production constraints. (2)

Soil fertility is a principal constraint to gowing field maize in the northern research

136

areas. This is followed by higher labor demands (relative to millet and sorghum) as well
as a higher susceptibility to droughts. (3) Natural predators, insects and high
labor/ equipment demands constrain the expansion of maize production in the southern
research areas. Similar wildlife and agicultural resource management issues have of
course existed for some time in countries such as Kenya, which depend heavily on rural
tourism. Finally, marketing problems pose potentially important further constraints.
They are discussed in Chapter V.

4.1.5. Recent Crop Mix Changes, Reasons Given by Farmers and Evidence

Relating to Farmer Decision-Making

Farmers in the survey areas by and large increased areas cultivated to crops in
1987 over 1986, as far as can be seen from qualitative responses, and the number of
households adding crops to their crop mixes outweighed the number reducing the extent
(or scope) of their crop mix (see GH, May 1987; and GH, March 1988 for changes over
the period 1984-6). The proportion of increases was somewhat biased in favor of food
crops, with maize production expanding noticeably (Table 4-11): On net, 7% or 14/203

TABLE 4-11 : ABSOLUTE AND RELATIVE CROP MIX CHANGES IN 1987 OVER 1986
Number and Percent of Households Responding

 

 

 

Number of Households Concerned As a Number

Percent of HHs

__QmILmL; W of Growing
Added Dropped Net Incrs. Reduc. Net firouere; Crop

Crop A D (A-D) I R (I-D) A—D I-O in 1986
Millet 6 14 -8 46 9 37 -7% 31% 120
Sorghum 9 11 -2 50 15 35 -2%. 27% 130
Home Maize O 2 -2 7 8 -1 -1% -l% 153
Field Maize 23 9 14 31 12 19 16% 22% 86
Rice 1 5 -4 9 2 7 -4%. 7% 99
Cotton 13 6 7 l4 1 13 . 11% 21% 62
Peanuts 17 7 10 75 22 55 6% 31% 180

 

Source: ISRA/MSU FSP Surveys, 1986/7
Note: Sample size is 203 households.

137
of the households-added field maize to their crop mix while 22% of the households
already gowing maize expanded the area allocated to that crop. At least in the
northern areas, where the floor price policy was in efiect during the 1987 marketing
season, this raises the question whether that policy or another reason motivated the
change.

The availability (cost) of peanut seed (37%), labor (21%), land12 (7%) and
fertilizer (4%) were common reasons advanced for changes in areas cultivated in the
north, along with the desire to meet household food (15%) and cash (3%) needs (see
GH, March 1988 for responses by crop).13 The producer floor price for cereals was not
evoked as a reason. In the southern areas, reasons given for increases in area cultiva-

V ted included the desire to meet household food (30%) and cash (14%) needs, increased
availability of labor and equipment (23%) and peanut (16%) and cereals (11%) seed.

All northern and most (93%) southern household heads indicated household food
needs were the first factor taken into account in deciding how much land to seed to
cereals. The expected relative price of cereals and cash crops was the second most
important factor for 39% (62 / 160) of farmers in both areas. Among southern farmers
giving a third factor, 43% (49 / 1 13) responded cereal seed available affected the area
seeded to cereals.

The evidence presented earlier in Table 3-13 (Chapter III) yields further insights

14 Most farmers are unlikely to

into how farmers approach crop production decisions.
invest in inputs such as fertilizer and agicultural equipment unless they are first able to

satisfy anticipated household food needs during the rainy season (GH, May 1987; Table

 

12This represents land previously lent to or borrowed from other farmers.
13 Similar responses were given for crop area decreases and changes in the crop mix.

14This section is extracted from GH, May 1987.

138
15 and Figures 2 and 3). Once these needs are met, most farmers in the north (31%)
would invest in peanut rather than cereals production. This supports the hypothesis that
many farmers, particularly in the arid north, would rather invest in additional peanut
seed before intensifying cereals production (see also Kelly, 1988a), and the hypothesized
peanut seed-labor linkage. In the south, hiring of labor constitutes an important use of
additional funds, followed by saving rather than investing of funds.

Thus, while formal-sector lenders tend to regard loans for food consumption as
unproductive, expanded food consumption during the hungy season can increase labor
productivity and availability. Some farmers reported they were unable to profit fully
from the good rains in the 1985-86 cropping season, because they lacked the energy to
expand area cropped and complete arduous cropping activities. By investing in higher
food consumption during the gowing season, farm households are able to put more land
area under cultivation and achieve higher yields through more careful land preparation
and weeding. They are also able to attract more laborers to work in collective cereals
fields, using additional food as a labor payment. Making loans available for food deficit
households to purchase cereals could be an essential element in breaking low income
rural households out of a cycle of low productivity, poverty and indebtedness.

Household heads in the south preferring to invest 15,000 FCFA in peanut seed
on average had larger cereals and lower peanut stocks than those who would have
bought food (Table 4-12). In the north, households heads preferring to buy peanut seed
with incremental funds had lower cereals (and peanut) stocks than households which
would have bought food. This lends further credence to the peanut seed-labor linkage.
In particular, the finding that some household heads would invest in peanut seed (and
fertilizer) before buying food suggests their decision-making is complex, and that trade-
offs are evaluated in an intertemporal context. The household head organizes and

manages cultivation of collective household gain fields (see Goetz and Diagana with

139
TABLE 4-12: HOUSEHOLD LEVEL STOCKS OF CEREALS AND PEANUTS,
STRATIFIEU DY FIRST INVESTMENT PREFERENCE FOR 15.000 FCFA
Stocks (July 1987) in Kilogramsa
North South

 

 

 

First Choice Cereals Peanuts Cereals Peanuts

Buy Peanuts 69 26 67 4
Buy Food 94 38 33 17

Source: ISRA/MSU FSP Surveys, 1986/7
a. Peanuts are per 1986 adult producer equivalent,
cereals per 1986 adult consumer equivalent.

 

 

 

 

 

Diallo, 1987; also Benoit-Cattin and Faye, 1982, and Chapter II), relying on family and
non-family labor. The more workers he can attract as dependents by providing peanut
seed, the more collective land can be put under cereals cultivation. By foregoing imme-
diate food consumption and buying peanut seed, household heads are indirectly investing
in geater food security for the coming year. This balance of productive investment and
consumption requires careful management. When gain supplies are not adequate to
meet food requirements, assets such as livestock or equipment must be liquidated, or
income needs to be tapped from other sources.
4.2. AUXILIARY INCOME-GENERATOR G STRATEGIES

Since agicultural households rely not only on revenues from crop production to
survive, it is important to identify at least the qualitative significance of non-crop
production activities and their role in providing non-production entitlements for food
access. Livestock ownership and off-farm income activities are described in this section.

4.2.1. Livestock

Table 4-13 suggests most households in the survey areas own some form of

livestock. Not surprisingly, members of the traditional livestock-herding ethnic goup

140
(Peulh or Fulani) are more likely to own larger cattle herds. Livestock transactions and

their role in assuring food security are examined in Section 5.1.3.

TABLE 4-13: OWNERSHIP OF LIVESTOCK IN THE STUDY AREAS
% of Households; excludes draft animals

 

 

 

 

 

 

 

 

 

 

 

Cattle Goats/Sheep Poultry
Number of

Animals North South North South North South
0 55 51 24 51 29 12

1-10 39 27 61 40 56 53

11-30 0 15 12 8 15 33
30+ 6 7 3 1 O 2
Peulh Other Peulh Other Peulh Other

0 41 67 43 34 16 21

1-10 38 28 44 56 54 53
11-30 13 1 11 7 29 25

30+ 8 4 2 3 1 1

 

Source: ISRA/MSU FSP Surveys, 1986/7

4.2.2. Off-Farm Income

Nearly two-thirds of the households sampled have at least one member who
pursued an off-farm activity during the 1986/87 agicultural season.15 The proportion of
households with off-farm activities is higher in the southern than the northern areas and
tends to be larger among households borrowing or not using agicultural draft
equipment. The data reveal a small net decline in the number of households in which
household heads pursued off-farm activities in 1987 relative to 1986. The decline may be

related to higher levels of coarse gain production in 1986, relative to 1985, which in turn

 

15This section is based on Goetz (June, 1988), "A Note on Off-Farm Activities and
Food Security in Southeastern Senegal“, which provides more details. An off-farm activity
is defined as any activity not related to own-crop production providing cash, including trade
in livestock but excluding disinvestment of livestock.

141
suggests that off-farm income is one important means of recuperating from crop failure
(1984-85 for these farmers). In other words, the household head may be forced to
pursue off-farm activities temporarily in order to allow his household to recuperate from
crop failures.

The household head was most frequently counted as the person pursuing the off-
farm activity (71% of the cases) followed by male (18%) and female relatives (11%). In
33% of the cases household heads engage in small-scale commerce (heneoene). Most of
the activities are carried out all year—possibly competing with own-crop production, see
Ch. II-or during the dry season, with some regional differences (Table 4-14). Obviously,
the volume of the off-farm activity may also vary seasonally.

TABLE 4-14: TIMING OF THE OFF-FARM ACTIVITY
Number and Percent of Households Responding

 

Study Region

 

 

Timing North South Total
All Year 22 32% 65 50% 87 44%
Dry Season 32 46% 55 42% 87 44%
Cash Crop Season 11 16% 2 2% 13 7%
Rainy Season/Harvest 4 6% 8 6% 12 6%
Total 69 100% 130 100% 199 100%

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)

Small-scale commerce (henebene) was the most frequently cited off-farm activity,
followed by artisinal and livestock-related pursuits (e.g., herding and trade, Table 4.15).
The blacksmith /bicycle repair activity was cited by only one northern farmer, suggesting
more specialization in the northern area, which is conceivably related to higher equip-
ment endowments. Only household heads engage in salaried labor and construction
(wells, huts), and most of the religious and livestock-related activities are carried out by

household heads. In contrast to finding from Mali (D’Agostino and Sundberg, 1990),

142

TABLE 4-15: OFF-FARM ACTIVITY FREQUENCY COUNTS
(All Household Members)
Number and Percent of Households

 

Study Region

 

 

Activity North South Total
Commerce/Beoefleoe 33 48% 41 32% 74 37%
Artisinal 12 17% 26 20% 38 19%
Livestock 8 12% 17 13% 25 13%
Religious/Marabout 3 4%. 13 10% 16 8%
Food Services 6 9% 8 6% 14 7%
Blacksmith/
Bicycle Repair 1 1% 8 6% 9 5%
Other 6 8% 17 13% 23 13%
Total 69 100% 130 100% 199 100%

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)
a. Includes construction (3%), salaried labor (3%), working
for other farmers (3%) and miscellaneous (4%).

these results suggest that off-farm activities pursued by farmers are not necessarily more
diverse and numerous in the more arid region (North); indeed, the institutional environ-
ment (better market integation, etc.) may compensate for a climatic deficit.

In three-quarters of the cases, the activity has been carried out for more than

three years (Table 4-16). The data imply more ”dynamism" in the northern areas,

TABLE 4-16: BEGINNING OF OFF-FARM ACTIVITY
Number and Percent of Households

 

Study Region

 

 

 

Beginning North South Total
1986/87 11 16% 5 4%, 16 8%
Two years ago 10 14% 6 5% 16 8%
Three years ago 7 10% 12 10% 19 10%
More than 3 yrs. ago 41 59% 103 82% 144 74%
Total 69 100% 126 100% 195 100%

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)

143
possibly due to opportunities ganted by recently liberalized cereals markets in 1986 / 87
combined with higher coarse gain surpluses. Commerce was cited most frequently as
having been started this past year, while construction and livestock-related activities are
more stable in the sense that they generally have been carried out for more than three
years.

Having described the multidimensional production behavior and possibilities of
agicultural households in southeastern Senegal, we now turn to estimating a cost
function to determine if the technology available to households is such that there are
economies of scope in producing food and cash crop within the same household.“ The
effects of ethnicity and off-farm incomes on the coarse gain marketing behavior of
households is examined in Chapter V.

4.3. COST FUNCTION ESTIMATION

43.1. Introduction

The purpose of this section is to estimate a household cost function, incorpora-
ting the multi-product nature of households, to determine the quantitative significance of
non-congested inputs. Labor may constitute one such input, since (for example) the
harvest periods of cash and food crops differ. From the estimated cost function, various
cost measures are calculated to examine where farmers are currently operating on their
cost curves. "Economies of scope", defined as the proportional cost reduction achiev-
able by producing food and cash crops in one rather than two specialized households, are
one such cost measure. Another is the economies of scale specific to one product in a
firm producing multiple products.

One of the explicit policy instruments the GOS has chosen to achieve its cereals

self-sufficiency objective is the substitution of coarse gains for peanut production in

 

16Livestock is excluded from the cost function owing to a lack of reliable production
data in 1986.

144

southeastern Senegal. The instrument is implemented by reducing sales of peanut seed
on credit and establishing a producer floor price for cereals. The question then arises,
why do most rural households in Senegal produce both food and cash crops? There are
two plausible answers to this question. 17

First, uncertain food markets create incentives for households to produce both
cash and food crops if they are to meet both cash and food needs of the household (Ch.
II). A second explanation involves the existence of a W or public input,
which makes it less costly to produce food and cash crops in one household rather than
in two specialized (separate) households. Non-congestion means that there is slack for
one or more inputs with regard to its use on a particular crop, and that gowing both
types of crops within the same firm (household) leads to a fuller utilization of the input
at any given point in time. If, alternatively, there were sizeable economies of scale in
gowing food and cash crops (separately), one could expect two or more households to
form an ageement or contract to specialize in the production of one or another of the
crops, and to trade outputs at harvest. For a small enough number of households, trans-
actions costs could be sufficiently small (i.e., close to zero) so that the gains from special-
ization exceed the costs of ageeing to and enforcing the contract. This form of com-
plete specialization across households, however, is not observed in Senegal. 18 One

reason may be that costs of enforcing contracts between households are high given the

uncertain environment.

 

17A third possible explanation, the peanut seed-labor linkage, was examined in Chapter
111.

18One can argue that specialization among different crops takes case githjn households
when coalitions are formed (see Ch. II). It is not clear whether this occurs because of gains
from specialization as such or because low profitability and market uncertainty, for example,
in food markets, leads non-family workers to almost exclusively gow cash crops. This may
imply that contract enforcement is less costly within than between households.

145

The remainder of this section is organized as follows. Section 4.3.2. more
precisely defines economies of scope and various cost measures associated with multiple
output firms. Section 4.3.3. discusses the translog cost function used for estimation pur-
poses, and section 4.3.4. presents the empirical results and interpretation of the finding.

43.2. Cost Measures for Multiproduct Finns

Following Panzar and Willig (1981), let Y=(Y1,...,Yn) represent the firm’s vector
of output levels produced out of the product set N= {1,...,n}, and Ys an n-dimensional
vector, the elements of which equal those of Y if the product is produced (i.e., for iES c
N) and 0 otherwise (i.e., for iGS). C(Y,W) measures the (minimal) cost of producing Y
given input prices W, while C(Ys,W) reflects the cost of producing output vector Ys’

Definition (Panzar and Willig, p.268-7): Let T= {T1,...,Tk} denote a non-trivial
partition of S g N. That is, ui'ri-s, TinT- .g for irj, Tim, and k> 1. There are econo-

J
mies of scope at Y3 and at factor prices W with respect to the partition T if

II MW

(1) f3(Y1-1.W) > C(Y3.W)-

1

Using the convention that Yi'(0""0’yi'0’"°0) denotes a non-negative orthogonal
vector, while Y represents a vector of all Yi vectors, an equivalent condition for the

existence of economies of scale is (suppressing W),

zicori) - cm
(2) (1308 > 0) o > 0
cm

 

This means it is cheaper to produce n products in one single firm rather than producing
only one distinct product in any of n separate firms, if the technology is such that it

exhibits economies of scope.

146

Product-specific economies of scale are calculated as
(3) PSE; ' AIC(Yi)/MC(Yi) " AIC(Yi)“¢5C(Y)/5)'i).
where MC is the marginal cost of producing Yi’ and AIC is average incremental cost, or
(4) Alcoa) - Ices/y,
where yi represents the (scalar) quantity of Yi produced, and IC(Yi) the incremental
cost of producing Yi' defined as
(5) 10m) - cm - C(Y - Y9.
Multiproduct scale economies can be calculated as
(6) MPSE - 1 - 2i81nC(Y)/61nyi.
Multiproduct scale economies exist if MPSE > 0, while the existence of product-specific
scale economies requires PSE > 1. These measures are discussed in more detail in
Baumol, Panzar and Willig (1982); Willig (1979); and Panzar and Willig (1977). For a
recent application to retail-level fertilizer firms see Akridge and Hertel (1986).

433. The Translog Cost Function

The translog cost function can serve as a local second-order approximation to any
arbitrary cost function (see e.g., Varian, 1984; or Binswanger, 1974 for a derivation of
the associated Taylor series expansion); with appropriate restrictions (see below), the
function satisfies linear homogeneity in input prices and with respect to outputs (in the
multiproduct case); and it is relatively parsimonious in terms of the parameters to be
estimated (Caves, Cristensen and Tretheway, 1980). One drawback of the function is
that it does not permit zero quantities since the regessors enter in logarithmic form. In
the estimation this problem is circumvented by replacing Yi-O with the value 0.10.19 To

facilitate the interpretation of the coefficients, and following common practice, the

 

19Caves et al. (1980, p.479) suggest using the Box-Cox metric
fi(Yi) =- (YiT - 1)/1 for 1 - 0 to circumvent this problem.

147
exogenous variables are normalized around their geometric means (Akridge and Hertel;
Kim and Ben-Zion, 1989).
Given the quantities,
Y =- a vector of n outputs;
W = a vector of p variable input prices;
F = a vector of r fixed factors; and total variable costs
C(Y,W,F) =- Xpwpxp,

where xp is the p-th variable input, the translog variable cost function is written as

(7) lnC(Y,W,F) =- ao + )chrnlnYn + Zpoanp + XrurlnFr +
VzfinzmAnmlnYnlnYm + EPZquqanpanq + ZrZsCrSInFIJnFs] +
21..2panlnYnanp + ZanEnrlnYnlnFr + ElperpranplnFr + e

As discussed in Brown, Caves and Christensen (1979, p.259), linear homogeneity
in factor prices is imposed with the restrictions
(8) )1po a 1; Xqu =- 0; for p-l,...,n and q--1,...,m

Xquq =- 0.

To ensure linear homogeneity in outputs and fixed factors, similar restrictions (some
redundant with (8)) are imposed on the following coefficients:
(9) 2nAnm = Enan g annr 3 0'

43.4. Estimation, Results and Discussion

Two outputs (Y1,Y2), one variable input price (W) and one fixed factor (F) were
used in the estimation (Table 4-17). The calculations for these variables are shown in
Appendix A-S. Since it is not possible to treat the outputs as strictly exogenous, a two-

stage estimation method was applied and predicted values of (Y1,Y2) were used in

estimating the cost function.

148

TABLE 4-17: COST FUNCTION STATISTICS FOR AGRICULTURAL MOUSEI-IOLDS

 

 

Standard b Geometric
Variablea Mean Deviation Minimum Maximum Mean
--------------------- in FCFA--------------------
Variable Costs 331,186 205,510 55,100 1,108,512 278,097
Food Crop Outputc 157,875 102,543 25,123 518,234 128,700
Cash Crodeutputc 210,034 219,097 .10 1,292,813 111,848
Nage Rate 396 83 257 534 387
Fixed Capital 30,955 34,679 .10 165,367 756

 

Source: ISRA/MSU FSP Surveys, 1986/7

Number of Observations - 145.00

a. All values expressed in FCFA; see Appendix A-5 for calculations.
b. Zero values recoded as 0.100 to compute logarithm.

c. Predicted values from another regression.

d. Per unit of four hours.

Coefficient estimates for the cost function are presented in Table 4-18. As
expected, both oi values are positive and statistically significant, indicating total variable
costs would increase with increases in the production of either crop (as expected).
Short-run average cost curves are U-shaped since the positive (albeit statisti-
cally insignificant at the 5% level) Afi values imply 6MCi/6yi>0 at the geometric means;
the positive B shows the cost function monotonically increases in input prices; while the
negative sign of 11 (although the parameter estimate is insignificant at the 5% level)
suggests households are in long-run equilibrium with respect to the use of fixed factors
(see also Akridge and Hertel, 1986). Hence for this sample the coefficient estimates of
Afi and u are of the expected sign but not statistically different from zero.

Producing both food and cash crops in the same household results in a 22.3%
cost saving relative to producing the same quantities in two separate (specialized)
households (Table 4-19). The average incremental cost of food production are less than
marginal costs of food production, while average incremental costs of cash crop produc-

tion exceed marginal costs of producing cash crops. This implies that for the average

TABLE 4-18:

149

OLS COEFFICIENT ESTIMATES FOR
THE TRANSLOC COST FUNCTION

 

 

 

 

Paramete

Variablea Estimate t-statistic
Intercept 12.387 61.050**
lnYl al .640 5.081**
lnYz oz .128 2.311*
lnP 8 1.029 5.419**
lnF p -.00277 -.074
lnYllnY1 A11 .138 .481
lnYllnYz A12 -.l46 -.493
lnY lnYz A22 .00784 .588
lnPInP B .134 .096
lnClnC C .00567 .481
lnYllnP 01 1.011 3.817**
lnYzlnP 02 -1.011 -3.817**
lnYllnF E1 .00593 .720
lnY lnF E2 -.00593 -.720
lnPTnF s -.0310 -.997
Number of observations 145
R Square .594
Adjusted R Square .561
Standard Error .397
F - 17.702**

-cash crop

a. Y -food crop output; Y
odtput; P-labor wage r

of fixed capital.
b. *ssignificant at 5%; ** at 1 % or

lower.

te; Favalue

150

TABLE 4-19: COST MEASURES FOR HOUSEHOLD CASH AID FOOD CROP PRODUCTION

Cost Data are Measured in FCFA

 

 

 

Product

Aggregate
Measure Symbol Food Cash Output
Stand-alone production cost C (Y) 180,413 121,407 301,820
Joint production cost (Y) 246,732
Economies of scope EOSY 22.3%
Incremental cost IC(Y ) 125,325 63,319
Average incremental cost AIC(I) 0.974 0.566
Marginal cost MC( Y I) 1.227 0.283
Product-specific economies PSE(I1) 0.794 2.000
Multiproduct scale economies MPSE 0.231

 

Sum: Tables 4-17 and 4-18.

household, reducing food production below and increasing cash crop production beyond

the geometric mean would reduce average incremental costs of production (see also the

estimates of the product-specific economies of scale). Conversely, inducing households

to substitute food for cash crop production, as planned in the NAP, would lead to higher

marginal costs of food production; equality of (observed) marginal revenue

and cost would in that case require a higher output price. It is of course plausible that

high transaction costs in risky markets mean that the observed market price of food does

not adequately reflect the insurance value of food. The PSE estimates also show

households are operating in the region of diseconomies of scale for food production, but

economies of scale for cash crop production. Finally, the existence of multiproduct

economies of scale (MPSE :- O.231 > 0) suggests a proportional increase in food and

cash crop production would entail a less-than proportional increase inn total variable

costs.

151
4.4. SUMMARY AND CONCLUSION

A variety of crop and non-crop production strategies—beyond own-food produc-
tion—are available to and pursued by households in southeastern Senegal. Not surpri-
singly, given the diversity of entitlements and resource ownership and use levels reported
in Chapter [[1, there is considerable diversity in terms of household crop production
levels. The use of animal draft technology appears to explain much of this difference,
particularly in the southern study areas. Policies designed to raise rural welfare will have
to take into account the diversity of resource ownership and the different food security
strategies that households pursue as a result. The following chapter, for example,
explores the potential for and the anticipated consequences of a floor price policy in
view of the finding that nearly one-quarter of the households sampled produced insuffi-
cient quantities of coarse gains (and rice) to last them until the subsequent harvest.
That chapter also attempts to discern how the different strategies contribute to house-
hold food security.

An examination of household cost functions yielded two important insights. First,
estimates of marginal and average cost of production suggest that households are already
producing "too much" food relative to cash crops, given observed relative prices and the
structure of their cost functions. Encouraging them to gow more food crops-and less
cash crops-would further distort this relationship. A plausible explanation given for this
phenomenon is that recorded market prices of food (and cash) crops do not adequately
reflect the insurance value to the household of having its own supply of food. Second,
there are certain non-congested inputs on rural farms, and economies of scope are
realized by producing both food and cash crops within the same household. These cost
saving to households (and the nation) would be lost if they were somehow forced or

induced to gow only food crops.

CHAPTER V
HOUSEHOLD MARKETING BEHAVIOR

Given the GOS’s objective of stimulating marketed surpluses of coarse gains in
southeastern Senegal, it is imperative to identify variables affecting the behavior of
households in coarse gain markets, and their opinions of planned market reforms. This
chapter describes the transactions behavior of households and examines post-production
issues (consumption preferences, transformation or processing constraints, etc.) in
Section 5.1., while Section 5.2 presents an improved method of estimating reduced form
marketed surplus equations to explain household behavior in coarse gain cash markets
in the presence of uncertainty and transactions costs. Section 5.3. summarizes the
research finding.
5.1. TRANSACTTONS BEHAVIOR OF HOUSEHOLDS

Using survey data on both manifest and potential behavior, tabular analysis is
used to describe households’ behavior in coarse gain cash markets; to determine the
food security situation of households before and after transfers; and to assess the uses
and quantitative importance of revenues from cash crop and livestock sales, as well as
off-farm activities, in assuring household food security. The data show households pur-
sue a variety of strategies beyond own-production of food to achieve food security, "a
fact that belies the conventional image of Sahelian peasants as autarkic agiculturalists"
(Reardon et al., 1988, p. 1065). Since consumption preferences, perceived storage and
processing problems may influence the cereals consumption mix of households, data are
also presented on this subject, with a special focus on maize.

The special role of rice imported unofficth from the Gambia is examined, and
an attempt is made to determine whether rice is a food security or a status good.

Finally, opinion questions about market reform and potential responses are used both to

152

153

verify manifest behavioral data, and to inform policy makers about households’ anticipa-

ted responses to output market reform.

5.1.1. Honseholds’ Coarse Grain Cash Market Position

As reported, many households failed to produce sufficient quantities of cereals to
last them until the subsequent harvest (Ch. IV). The shortfall in production induced as

many as 30% of the households to enter coarse gain markets as net buyers in the 10

month period following the 1986 harvest (Table 5-1). Nearly 40% of the households

failed to participate in coarse gain cash markets altogether.

I

In contrast, most house-

holds (79%) purchased rice, most of which was imported privately from the Gambia

(originating in Asia). Also, 42% of the households with food production shortfalls did

not enter cash markets to purchase coarse gains. As argued in Section 5.2., this

phenomenon is largely explained by prohibitive (transactions) costs, especially in the

more isolated southern areas, involved in articulating supply and demand of individual

TABLE 5-1: COARSE BRAIN MARKET PARTICIPATION PROFILE

(Standard Error)

 

 

 

 

 

 

0 Percent of
Not in the Production
Study Region Net Buyer Market Net Seller Marketed
and Rainfall % kgs % % kgs %
North/550mm 20 243 21 59 339 8
(95) (57)
South/900mm 38 332 53 9 279 2
(67) (118)
Total 32 308 40 29 327 5
(55) (51)

 

Source: ISRA/MSU Food Security Project Surveys (1986/87).

 

1In one triad (Sare Yoba Diega), where livestock sales are important, virtually no cash
transactions were observed in coarse gain markets.

154

households for coarse gains through cash market transactions.2

On average only a
small percentage (5% for the overall sample) of total production found its way into rural
cash markets.

Market sales were concentrated in a small proportion of households, with 50%,
70% and 80% of total sales by volume effected by 7, 11 and 15% of all households. The
short run implication of these finding for food price policy is obvious. An effective
higher cereals floor price would have only narrow distributional benefits, and would
likely impose substantial costs on rural households relying on markets to purchase food.

Table 5-2 presents the unambiguous buy-sell market position of households
during the transaction period surveyed, and also displays information about selected
household characteristics of the various goups. In particular, instead of looking at only
net positions, households are divided into those that only bought (26% of the sample),
only sold (24%), did not participate in coarse gain markets (40%), and those that both
bought and sold coarse gains (10%). Thus from the goup of net buyers (32% of the
sample) and net sellers (29%) identified in Table 5-1 we can "extract" those that orgy
bought, and those that mth bought and sold but on net were nevertheless net buyers.
The results show households only buying coarse gains during the 1987 dry season
(period a, see Ch. 11) lived in market areas (triads) where they on average faced a 10
FCFA/ kg higher price than did selling households in their market areas. This price was
3 FCFA we the floor price, which raises the possibility that buying households already
face higher buying prices than do selling households, and that they may at least in the
short run have difficulty responding to a higher floor price offered as a production
incentive. In fact, it is not known whether these buying households also expected to face

higher prices, and the results have to be viewed with caution since only aggegate prices

 

2It is nevertheless surprising that even transactions within villages were absent in the
Sare Yoba Diega triad.

 

155

TABLE 5-2: FARM HOUSEHOLD CHARACTERISTICS AND COARSE GRAINa
CASH MARKET POSITION: 1986/7; 10 MONTHS
(Standard Errors)

 

 

M P i i n
Household Buy Not in Sell Se
Characteristic Only Market Only a Buy
Households (%) 26 40 24 10
Quantity of per capita CG
sold [bought]; in kgs [42] -- 64 8
(7) (16) (10)
Average coarse grain
market price; in FCFA/kg 73 66 63 64
(3) (2) (1) (2)
Average rice market
price; in FCFA/kg 169 168 149 152
(2) (2) (3) (4)
Percent Fully
equippedc 40 40 66 71
(8) (7) (8) (13)
Value of equipment per
active worker; in FCFA 2160 2870 3130 3300

(447) (438) (440) (652)

Fulani Ethnic Group;
in percentc 76 so 46 36
(7) (7) (9) (13)
Mechanical transformatign

technology available 5 9 26 0
(4) (4) (7) (0)
Per capita Coarse Grains
produced/1986: in kgs 145 240 374 247
(18) (26) (43) (30)
Per capita cash crop
produced/1986; in '000 FCFA 17 36 29 36
(3) (5) (5) (8)
Coarse grains stocks/capita
as of July 1987; in kgs 22 46 64 45
d (4) m (11) (9)
Estimated utilization
of cereals/capita; kgs 181 166 190 205
(13) (12) (15) (21)
1985 CG production
lasted < 9 months [11° 53 51 29 29
(8) (7) (8) (13)
1987 C6 area seeded < HH
consumption needs [‘%]c 66 48 11 27
(8) (7) (5) (12)

 

Source: ISRA/MSU FSP Surveys. 1986/7

Notes: a. Coarse Grains are millet. sorghum and maize
b. Number of households is 39; 61; 36: 15. respectively.
c. Percentages refer to the percent of households per stratum.
d. See Table 5-4. note (6). for calculations.

156
from five market areas are available in this study. Nevertheless, the finding tend to
support the assertion that some households in southeastern Senegal will have difficulty

responding to floor price incentives in the short run.3

More applied research that
generates transaction-level prices would be required to confirm or refute this assertion.
Households selling coarse gains were also exposed to lower rice prices than buying
households (on average 20 FCFA / kg lower), suggesting they were selling coarse gains to
purchase rice as a preferred staple.

Examining the information in Table 5-2 about other household characteristics, it
is not clear whether or not draft equipment ownership, or the value of equipment per
active worker affects a household’s market position. The data do suggest, however, that
selling households are somewhat more likely to be fully equipped and to own more
equipment per active worker. Since fully equipped households gow more cash crops
(Ch. IV), they may not have to enter coarse gain cash markets to meet cash needs. In
contrast, households borrowing equipment may enter these markets to repay debts from
borrowing equipment.

Members of the Fulani ethnic goup are more likely to buy coarse gains,
indicating that traditional livestock herders were liquidating livestock holding to
purchase coarse gains (Section 5.13.). Households selling coarse gains are more likely
to live in villages with access to mechanical coarse gain processing facilities (within a 10
kilometers radius). Availability of this threshing machines, which may reduce the cost of
threshing coarse gains relative to manual methods, is potentially important since cereals

 

3Agicultural credit is limited so that households are generally unable to capitalize
higher future food prices into current period inputs; since buying households also rely on
markets to buy food, a higher buying price could set off a vicious circle from which it is
difficult to escape. Private credit taken out during the gowing season is used primarily for
short-term consumption purposes, and thus serving to increase labor productivity, but not
to increase non-labor inputs. Thus higher coarse gain prices would make it more expensive
to increase production because of the increased cost of labor paid for in the form of food.

157
enter cash markets in tlnreshed form (early harvested "geen' maize, eaten on the husk
during the hungy season is an exception).4

Results in Table 5-2 also show that the average quantity of coarse gains
produced per capita was lowest among households only buying coarse gains. Similarly,
ending stocks of coarse gains, including seed, tended to be lowest for households only
buying coarse gains. The estimated per capita availability of all cereals (including rice)
was similar for buying and selling households, but lower for households not participating
in the market.

The last two rows in Table 5-2 provide further potential insights into the ability
of households to adjust to a floor price incentive. Households which sell only-i.e., those
able to benefit from a higher floor price if and when it was effective-were apparently
able to increase their area seeded to coarse gains over the period 1985-87, with the
proportion of households not producing enough food to meet their consumption needs in
that category dropping from 29 to 11% (in contrast, these proportions remained fairly
stable among households in the other market position categories). This may reflect
investments in coarse gain production facilitated by the floor price, which prevailed to
different degees of effectiveness in the northern areas (the goup of sellers on average
nevertheless faced a price 7 FCFA/kg below the official price of 70 FCFA/kg in 1987).

An often-mentioned phenomenon in African agiculture is the selling of food at
depressed prices soon after harvest to generate cash, which is used in turn to pay off
debts incurred during the hungy season and / or to meet other cash needs. More often
than not, households following this strategy are said to have to buy back food later on in
the season (perhaps using income earned during the dry season) at a higher price which

reflects storage costs. The results shown in Table 5-2 make it clear that this

 

4See also Holtzrnan (1989).

158
phenomenon pertains at most to only a small goup of households. The 10% of the
sample (15 households) that both sold and bought coarse gains during the survey
period, were scrutinized to determine whether they were pursuing this strategy of
”forced” sales, or whether they were responding to NAP incentives of liberalized cereals
trade and becoming cereals traders.

The results presented in Appendix A-6 generally suggest that the hypothesis of
"forced" sales, even for this relatively small goup of households, is not confirmed in
southeastern Senegal. Only about one-third of this goup of households do indeed
appear to sell cereals soon after harvest and then buy back more cereals later in the
season. To delve further into this question, the 15 households in the buy-and-sell
category were split into net buyers and net sellers (see the table in Appendix A-6). The
general conclusion from this table is that these households are (on average) quite well
off in terms of the indicators listed (equipment value per worker, value of cash crops
produced and the estimated availability of cereals per capita) relative to the households
in the "buy only“ category. Of the households that both buy and sell coarse gains, net
sellers tend to be more equipped (they erhibit the highest mean of all goups), produce
more food and cash crops, and are less likely to be of the Fulani ethnic goup in
comparison to the net buyers. Nevertheless, the estimated cereals availability is virtually
the same in both cases.

5.1.2. Food Availability Before and After Transfers; Stocks

Since many households in southeastern Senegal failed to achieve food production
sufficiency in 1986, this section investigates how households assured food consumption

security.5 Table 5-3 shows the relationship between the food production security of a

 

5 Food consumption security is achieved through own-production of cereals and other
entitlements permitting access to food (following Sen, 1987). As suggested in Ch. II, food
consumption and food production security are two distinct concepts.

159
household and its net coarse gain market position. Buying households are more likely
to have low production per capita, but this relationship is not always clear-cut (i.e.,
households may also rely on barter trade and gifts to make up the food production
deficit). The table also shows that some households with medium and high levels of
production per capita become buyers of food (in other words, food buying households

are not exclusively low producirng households).

TABLE 5-3: RELATIONSHIP BETHEEN HOUSEHOLD FOOD PRODUCTION
SUFFICIENCY STATUS AND NET COARSE BRAINS MARKET POSITION

 

Kilograms of all Cereals
produced per Household Consumer

 

 

Household

Coarse Percent Low Medium High

Grain Market of 0-160 161-300 301+ Total
Position Householdsa kgs kgs kgs

Buy Only 26% 53% 37% 11% 100%
Not in Mkt. 40%» 36% 29% 35% 100%
Sell Only 24% 9% 43% 49% 100%
Sell & Buy 10% 14% 57% 29% 100%
Percent of

Householdsa 100% 32% 37% 31%. 100%

 

Source: ISRA/MSU FSP Surveys, 1986/7
a. Total number of households is 142.

Crop disposal and acquisition behavior in the post-harvest period is further
examined below along with household heads’ stated reasons for selling cash and food
crops. Anticipated short run consequences and problems of implementing an effective
floor price are also identified.

The coarse gains production deficit in the southern areas was on average made
up for through purchases of coarse gains and—mostly Gambiano-rice (Table 54).

Although equipment borrowers in the north on average produced and sold more cereals

160

TABLE 5-4: HOUSEHOLD CEREALS TRANSACTIONS. RELATED TO REGION
AND EQUIPMENT USE, 1986/7 (IO-MONTHS)

 

 

1986 fierce]; Iraneectjonc oer 5:95
Production Ne; Sale; Net Stock Estim.
Study of all Coarse Other July Utili-
Region Cereals/AFC Grains Rice Out 1987 zation
and Eqpmt- (1) (2) (3) (4) (5) (6)
North
Borrower 376 26 -10 52 119 189
Fully.Eq. 283 16 -19 34 65 189
South
Non-User 158 —20 -28 11 28 167
Borrower 188 ~15 -20 19 28 175
Fully Eq. 195 -23 -10 24 30 174
Avg.Smpl. 227 -4 -19 26 45 181

 

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)

Notes:

a. Data are for a subset of 109 valid households; survey period is
harvest (October/November) 1986 through July 1987.

b. Adult-equivalent consumers are computed using the following weights:
adult 15 years + - 1.0, child 5-14 - 0.5, and infant 0-4 - 0.25. Nawe-
taans and their family members - 0.4 (4.8 months/year) adult consumer
equivalents.

(1) Includes coarse grains and rice. Coarse grains are converted into
consumable product equivalents assuming a 78 % transformation rate for
millet, sorghum and maize (FAO, 1984). Milled rice-equivalents are ob-
tained assuming a 65 % conversion rate between paddy and dehulled rice.
For the sample, there are .75 adult equivalent consumers per capita
(i.e., head or person). Consequently, when pro-rated to 12 months, the
above estimated utilization data are roughly comparable with (unshelled)
coarse grain consumption per capita per annum (see also Note (6)).
(2) and (3) Net sales are sales minus purchases.

(4) Barter, gifts and loans and (5) include CPE millet-sorghum-maize and
r ce.
(6) Calculated as production - all net transfers out - stocks as of July
1987 - estimated utilization over IO-months. The calculation assumes no
carry-over stocks from 1985 and does not account for harvest and storage
losses. See also Note (1).

161
per capita (adult-consumer-equivalent) than did fully equipped households, the means
have to be interpreted with caution owing to large standard errors (also, less than 30%
of the northern sample were in this equipment category). In the southern area, conver-
sion of equipment non-users to users would appear to increase cereals production per
capita but not alter average quantities of coarse gains purchased significantly. At the
same time it is interesting to note the relationship between rice purchases and equip-
ment classification in the southern area. Northern households that borrowed equipment
exhibited the largest ending stocks (in July) and on average achieved an estimated
utilization level identical to that of fully equipped households.

Average estimated revenues from sales of cash crops (FCFA 30,200/AEC in the
north, FCFA 16,900 /AEC in the south) in principle provided food production-deficit
households with enough cash income to become food consumption-secure by purchasing
cereals. However, households must allocate cash between food end non-food expendi-
tures, which together are estimated at FCFA 30,000/ABC for a typical household6.
Sales of livestock, barter of goods and services and income from non-farm activities
provided additional means of obtaining food (see Section 5.1.3).

Ten percent of the southern households entered cash markets to acquire more
than 50% of their estimated cereals disappearance (both rice and coarse gains).7 Forty
percent of the southern households acquired 20% or more of their disappearance in this
manner. In comparison, only 10% of the northern households are estimated to have
relied on cash market transactions to acquire 20% or more of their disappearance during

the 10-month post-harvest survey period (through July, 1987).

 

GBased on a survey of village chiefs in the sample. Excludes rural taxes and large
expenditure items such as marriages.

7Availability estimates assume no storage losses and zero beginning stocks. Ending
stocks (July) are included in the calculations.

162

Table 5-5 provides averages for household crop transactions and the estimated
disappearance of coarse gains according to households’ food production status. It
appears that households in the low-producing category produce proportionately more
rice than do households in the high-producing category. This could be related to labor
constraints within the household (as mentioned earlier). Households in the medium
own-production category on average sold a small amount of coarse gains, while rice
purchases are on average similar for each category. Per ABC cash crop sales (and
production) tend to increase from low to high food-producing households, and house-
holds in the low category tend to sell a smaller proportion of their peanuts. It is
conceivable that these households must retain their own peanut seed since they can not
(or can perhaps no longer) rely on the parastatal (SONACOS) to provide them with
seed.8 Finally, the estimated disappearance (consumption) of all cereals per capita
increases significantly across the three categories. This implies access to cereals via own
production is important in the survey areas. Viewed alternatively, relaxing production
constraints is likely to significantly relax consumption constraints. In terms of the cash
crop-food crop ”debate" it is also noteworthy that households in the high food own-
production category sell (and gow) significantly more cash crops than do households in
the other two categories. This in turn suggests that the expansion of food crop produc-
tion over time goes hand in hand with increased cash crop production (and vice versa).

Although rural head taxes, estimated at 1000 FCFA per active worker regardless
of age and gender, are small relative to the average value of all (cash and food) crops

sold, survey results reported in Table 5-6 show they are the most important first reason

 

8This may be related to past debt repayment problems and / or lack of collateral in these
households.

 

163

TABLE 5-5: HOUSEHOLD MARKETING AND ESTIMATED CONSUMPTION
CHARACTERISTICS (PER AEC) ASSOCIATED HITH HOUSEHOLDS’
FOOD PRODUCTION SUFFICIENCY STATUS
(Standard Errors)

 

 

 

Kilograms of all Cereals Total
Produced per Household Consumer
Per AEC for Low Medium High
all Continuous 0-160 161-300 301+
Variables kgs kgs kgs
Percent of HHs 32%. 37% 31% 100%
Cereals Output
in kgs 98 231 487 268
(6) (6) (32) (17)
Percent of which
is Rice (South) 24% 18% 13% 20%
(5) (2) (4) (2)
Net CG Sales &
[Purchases], kgs [22] 4 35 5
(6) (6) (14) (5)
Rice Purchases
(Net) in kgs 20 16 21 19
(5) (2) (4) (2)
Cash Crop Sales
(Net) in FCFAa 12558 22291 38298 24260
(1783) (2793) (5313) (2212)
Pcnt.of Peanut
Production Sold 58 73 69 67
(5) (3) (3) (2)
Kg Disappearance
of all cereals 125 182 260 181
{10 Months} (10) (7) (12) (7)

 

Source: ISRA/MSU FSP Surveys, 1986/7
Note: See Notes to Table 5-4 for the interpretation of
AECs and disappearance (or utilization) estimates.
a. Excludes 4 households claiming to be net buyers of
cash crops.

164

TABLE 5-6: FIRST REASON FOR CROP SALES,
1986/7; 10 MONTHS

 

Percent of Responses

 

 

b Total
First Reason Food Cropa Cash Crop N %
Pay Taxes 8 36 57 29
Social/Ceremony 32 19 44 22
Reimburse Debts 12 22 39 20
Buy Consumer Goods 14 10 22 11
Buy Cereals (Rice) 18 5 17 9
Agric. Investment 6 2 6 3
Other Reason 10 5 12 6

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)
Notes: (a)-millet, sorghum, maize; (b)-peanuts and cotton.
Based on the first reason advanced by respondents.

given by household heads for selling peanuts and cotton.9 This suggests that rural taxes
are one reason for gowirng cash crops, possibly detracting from the production of cereals
which historically have not had reliable market outlets. If one excludes 12 (7%) house-
holds claiming not to have sold any crops during the survey period, the value of taxes
paid exceeded 20% of the goss value of all crops sold in 10% of the households. Sales
of food for ”social and ceremonial" reasons suggest some households use cereals to meet
unexpected cash needs (deaths, medical bills, helping relatives), but do not plug to sell
cereals.

Table 5-7 shows the incidence of taxes on households according to their food
production security and coarse gain market participation status. Head taxes represen-
ted a larger percentage of the value of coarse gain transactions, cash crop sales and
total crop production in households producing low amounts of cereals. The tax burden

on average exceeded 75 % of the amount spent on food for households only buying food.

 

9Farmers sell peanuts at government-subsidized prices. This raises the question whether
both taxes and subsidies should be reduced, and what the consequences would be for
different goups of farmers.

165

TABLE 5-7: INCIDENCE OF TAXES ON HOUSEHOLDS HITH DIFFERENT
FOOD PRODUCTION SUFFICIENCV STATUS AND MARKET POSITION

(Standard Error)

 

 

 

 

 

 

Kilograms of all Cereals Total
produced per Consumer
Head Taxes as Low Medium High
a Percent of 0-160 161-300 301+
Net CG Sales
[Purchases] [27] [3] 16 [5]
(18) (32) (19) (15)
Cash Crop Salesa 15 9 3 9
(6) (2) (0) (2)
Total Crop b
Production 5 2 1 2
(1) (0) (0) (0)
Coarse Grain Market Position Total
Buy Sell Buy &
Only Only Sell
Net CG Sales
[Purchases] [77] 71 4 [5]
(19) (15) (59) (16)

 

Source: ISRA/MSU FSP Surveys, 1986/7
Average excludes 4 households claiming to be
net buyers of cash crops (i. e. ,peanuts).
Peanuts are valued at 90 FCFA/kg, cotton at
100 FCFA/kg (both are official prices).

b. Coarse grains are valued at prevailing
average market prices, rice at 150 FCFA/kg.

166

Virtually all cash crop sales were completed prior to March 1987. In the nor-
thern areas farmers on average sold a larger quantity of cereals during the period
harvest 1986-February 1987 (10 Kg/AEC) than during March 1987-July 1987 (6
Kg/AEC). Yet average cereals prices observed on rural markets were 10 FCFA/ kg
higher in the latter period (67 vs 57 FCFA/kg); see also Figures 5-1 and 5-2.10 A likely
explanation for this sales pattern is that some farmers need cash before the official
cereals and peanut marketing seasons open and sell cereals during these periods of
relatively low prices.11 Enumerator estimates of total cereals quantities moving through

the two northern markets12

generally confirm this marketing pattern for millet and
maize but not for sorghum (see Figures 5-1 and 5-3), which came from large sorghum
producers outside the sample during February through July. .13 There is an inverse
relationship between sales of cereals, distributed bi-modally around November through
January and May, and peanut sales which peak during J anuary-March.

The CSA had no gounds for purchasing cereals in the southern weekly markets,

where households were on average net buyers of cereals, regardless of equipment use

levels. 14 Purchased rice contributed significantly to the estimated total consumption of

 

10See Goetz, April 1988, for price series of other crops.

11One function of the Food Security Commissariat (CSA) is to buy cereals from
farmers at the official price (70 FCFA/ kg). For logistical and other reasons it began
purchasing cereals in the research areas in January 1987, roughly two months after the
cereals harvest.

12Estimates of this kind are characterized by measurement problems and the data
should be viewed as indicative of seasonal patterns.

13 Some of the sorghum was attracted from the Gambia by the floor price.

14There were some sales by surplus producers, but most of these took place directly
between farmers. One exception is the Medina Yoro Foulah area (outside the sample)
which produced surpluses of cereals for export from the area; here the CSA failed to
intervene because prices were close to the official floor price and because Gambian coarse
gains were brought to the markets.

 

167

 

 

 

FIGURE 5-1: PRICE. AND ESTIMATED QUANTITIES OF CEREAL. COLD
BY FARMER8 IN NOIAPTO ARRONOISSEMENT or KOUNOHEUL)
OCTOBER 1980 - JULY 1987
°° .. ’°°°° LEGEND
,01’ If. 4\ ’{j - — cu ’"°.
. "aeooo
/- \/ \ ,_ ...- Mlllot
...... Mono
Ouch. M".
A
3‘ E Ouch. Moi.

 

 

 

V
\
\
\
\
\
x
N
\
\
\
\
\
\
N
\
\
\
\
§
.\

m. I'M/MN 7" Cum. 1.00/7

 

 

FIGURE 5'2: PRODUCER CEREAL. PRICES IN THE WEEKLY

 

 

 

 

 

 

 

 

 

 

 

 

 

  

    

 

   

MARKET OF NDOOA MEACAR. MAKA: AUGUST 1986 - 1987
‘°° \ LEGEND
\\ _ on one.
“’1’ \\ ...... Millet
I\ ...... Sorghum
no" “3 . _- Moizo
if I
g \\
b 700 \I ‘._ ___ fl.
. '-, .- ~ /‘\\
'°*’ l\ '- i V \\
\ \1. //’ \\
so? I \.‘~..[’ ‘9
I.”
U-—’ \
‘0 r
Aug Sop Oct Nov Doo Jon Foo Mor Apr May Jun Jul Aug
Month
Mn I'M/MN "P Dom IMO/7
HOURE 5-3: ESTIMATED QUANTITIES DP CEREALS SOLD DY
FARMERS PER MARKET DAY IN NDOOA DADACAR (MAKA-TAMDA)
”°°° LEGEND
7//z MIIIot
aeooov - Sorghum
m Molt.
Emit
noooo»
10000"
eooo< " a
E ' 2
r g F g
o, a s. I 2
Aug Sop Oct Nov Doo JD? Fob Mar A99 Moy Jun Jul Aug
Month

M'

ISRA/MW no law. IODO/7

 

 

168

rice in these southern households despite better-than-average rainfall. An estimated
13% of the rice consumed during the period Harvest 1986 - February 1987 was purcha-
sed.15 During March-July 1987, this percentage is estimated to have risen to 72%.
Equivalent percentages for coarse grain purchases in the southern areas during these two
periods are 8 and 18%, respectively.

5.1.3. Cash Crop Sales, Livestock Sales and Oil-Farm Activities

As suggested above, households have both food and cash needs, and cash crop
sales—aside from serving to repay production credits to SONACOS or SODEFITEX-
play an important role in meeting cash needs. On average 65 % of the peanuts harvested
were sold during the 10-month survey period (Table 5-5), the proportion being 100% in
the case of cotton.

Reasons given for livestock sales three months prior to and during the 1986 rainy
season (see Goetz with Dieng, March 1987, p. 29), suggest livestock serves an important

16 In the three month period

role in assuring food security during the hungry season.
preceding the hungry season, 28% of all livestock transactions were carried out to
acquire cereals. In the hungry season that percentage rose to nearly 50% of all livestock
transactions. Hence the motivation for selling livestock changes seasonally.

Sales of small ruminants and poultry during the 1987 dry season (approximately
October/November 1986 through July 1987) appear to serve an important auxiliary role
in meeting the cash needs of some households (Table 5-8). Households in the ”low“

producing category on average sold enough small ruminants and poultry to acquire 24

 

15 Calculations assume no carry-over stocks of rice from 1985 production, but include
ending stocks.

16The importance of borrowing (private sector) cash to acquire food during the hungry
season is reported in OH (May 1987, pp. 27 ff.): most loans were taken out during August
(the peak of the hungry season), and in 57% of all cases the loans were used to purchase
food.

169

TABLE 5-8: LIVESTOCK TRANSACTIONS AND
FOOD PRODUCTION SUFFICIENCY STATUS
(Standard Error)

 

Kilograms of all Cereals
produced per Consumer

 

 

Livestock Low Medium High
Transaction 0-160 161-300 301+
W
and Poultry:
Net FCFA Sale
[Purchase] 1696 [2003] 690
(1568) (1112) (1940)
a: Net Sellersb 27% 32% 32%
% Net Buyers 24% 42% 39%
l
% Net Sellers 16% 21% 11%
% Net Buyers 4% 4% 11%
flames
% Net Sellers 2% 2% 11%
% Net Buyers 0% 6% 7%
% Net Sellers 0% 0% 0%
% Net Buyers 2%. 9% 14%

 

Sourggz ISRA/MSU FSP Surveys, 1986/7
a. Includes sheep, goats and poultry.
b. Percentages add up by column.
For example, 80% of the households
in the low food production
category neither sold nor bought
cattle.

170
kgs of coarse grains per adult-equivalent consumer (using the official price of 70
FCFA / kg). Nevertheless, 27% of the households in this category were also able to
invest in this type of livestock. It is important to separate transactions involving small
ruminants, poultry and cattle on the one hand, and net investments in draft animals
(oxen, horses and donkeys), on the other. To this end Table 5.8 also shows the percen-
tages of households investing in and selling ofl’ difl'erent types of draft animals. It is
noteworthy that a larger proportion of farmers were sellers than buyers of cattle. This
appears to reflect exports of cattle from the research area to urban consumption areas
(Ziguinchor, Kaolack and Dakar). In contrast, donkeys were introduced into the
research areas by traders from the Peanut Basin.

A similar picture emerges when livestock transactions are stratified by the coarse
grain market participation status of the household (Table 5-9). Households only buying
coarse grains on average sold off an amount of small ruminants and poultry correspon-
ding to 10 kgs of coarse grains per adult-equivalent consumer. Surprisingly, the propor-
tion of net buyers (34%) exceeded that of the net sellers (24%),however. At the same
time, 21% of the households in this category did sell off cattle during the survey period.

An examination of the net per capita value of transactions involving all types of
animals and livestock (including draft animals) reveals the following: southern house-
holds, Fulani households, households owning cattle and households without off-farm
activities on average sold off more livestock than households without these charac-
teristics (Table 5-10). As in the two preceding tables, the large standard errors suggest,
however, that the averages have to be interpreted with caution. The data thus indicate
that livestock plays an important role in the economic affairs of households in south-
eastern Senegal, and this sector should not be ignored when rural development policies

are designed. Binswanger et al. maintain livestock is the only major means of long run

171

TABLE 5-9: LIVESTOCK TRANSACTIONS AND
COARSE BRAIN MARKET PARTICIPATION
(Standard Error)

 

CG Market Participation

 

 

Livestock Buy Not in Sell Buy &
Transaction Only Market Only Sell
Sm in n
and Poultry:
Net FCFA Sale
[Purchase] 693 [541] [2280] 5905

(1211) (1561) (2033) (2146)

% Net Sellers 24% 31% 43%. 64%
% Net Buyers 34% 35% 26% 14%

i l

% Net Sellers 21% 16% 11% 14%

% Net Buyers 3% 7% 3% 14%
flgrses

% Net Sellers 0 2% 14%» 7%

% Net Buyers 3% 7% 3% _ 0
0.911311:

% Net Sellers 0 0 O 0

% Net Buyers 13% 6% 3% 21%

 

Sggggg: ISRA/MSU FSP Surveys, 1986/7
a. In FCFA per adult equivalent consumer.
Net sales are positive.

172

TABLE 5-10: PER CAPITA NET LIVESTOCK TRANSACTIONS
BY HOUSEHOLD CHARACTERISTIC
(Standard Error)

 

 

 

 

 

Off-farm

Region Ethnicity Cattle Owner Activity

North South Fulani Other Yes No Yes No
----------------------------- in FCFA---------------------------
837 2367 4830 [2601] 4017 683 267 3879

(1748) (2513) (2076) (1794) (2085) (2016) (1987) (2078)

Sooroo: ISRA/MSU FSP Surveys, 1986/7
Note: Numbers in [] are net purchases

 

capital accumulation in African agriculture with its relatively land-abundant and clima-
tically uncertain environments.

The "off-farm activity sector," encompassing many different economic activities, is
hypothesized to be of an importance similar to that of the livestock sector. Household
heads were asked who in their household pursues what type of off-farm activity, during
which period(s) of the year, how the proceeds were used, and how much they contri-

buted to the food needs of the household. Results shown in Table 5-11 suggest that

TABLE 5-11: OFF-FARM INCOME BY FOOD PRODUCTION SUFFICIENCY STATUS
(Standard Error)

 

Kilograms of all Cereals
produced per Consumer

 

 

Per AEC Low Medium High
Variable 0-160 161-300 301+
----% of households----
Someone in NH
Pursues an Off- 67 6O 50
Farm Activity (7) (7) (8)
Household Head 64 53 43
Pursues Activity (7) (7) (8)

 

Sooroo: ISRA/MSU FSP Surveys, 1986/7

173
households in the "low" production category were somewhat more likely to pursue an off-
farm activity relative to households in the other categories. Also, the household hood is
more likely to be involved in the off-farm activity in the case of households in the "low"
category.

Southern households are more likely to use proceeds from off-farm activities to
acquire food and these proceeds are more likely to contribute more than half of the
household’s food during the period in which activities are carried out (Table 5-12). In
cases where household heads pursue the activity proceeds are frequently used mainly or

exclusively to acquire food.

TABLE 5-12: USES OF PROCEEDS FROM OFF-FARM ACTIVITIES
AND CONTRIBUTION TO FOOD AVAILABILITY

North South Total

 

 

Used only for food 6 9% 16 12% 22 11%
Used mainly for food 29 43% 81 62% 110 56%

Used mainly
otherwise 23 34%. 27 21% 50 25%
Not used for food 10 15% 6 5% 16 8%
TOTAL 68 100% 130 100% 198 100%

 

More than half 9 20% 47 54% 56 42%

Half 6 13% 8 9% 14 11%
Less than half 30 67%. 32 37% 62 47%

TOTAL 45 100% 87 100% 132 100%

Soogog: ISRA/MSU FSP Surveys, 1986/7

Note: Data are for the subset of (132) households with at
least one member pursuing an off-farm activity.
”Use of proceeds” data are at the level of indivi-
dual members pursuing an activity.

 

In the northern areas, households without off-farm activities sold nearly twice the
amount of coarse gains per adult consumer equivalent as households with such activi-

ties. They also tended to have a marginally higher disappearance of cereals per capita.

174
In the southern areas, households with off-farm activities purchased two times the
quantity of rice as households without such activities, and apparently experienced a
slightly better cereals situation than did households without off-farm activities.

Ofi-farm activities are therefore also important in achieving household food
security, and they may contribute to agicultural production by enabling investments in
agicultural inputs. We will see off-farm activities increase the probability of coarse
gain purchasing and sales. This is hypothesized to reflect both the increased willingness
and ability of households to participate in coarse gain markets, and the increased
market information available to these households.

5.1.4. Consumption Preferences; Storage and Processing Issues

To gauge farmers’ preferences with respect to different cereals, and to anticipate
potential demand-side problems of expanded maize production complementing results
reported in Ch. IV, questionnaires were designed to elicit information on how, and why,
farmers would chose among rice, millet, sorghum and maize if they had a choice. They
were also asked about difficulties experienced with maize (and rice) storage and trans-
formation. This section presents results from these questions, while Section 5.1.5. and
5.2.5. examine changes in rice consumption in 1986/87 over 1985/86 and attempts to
determine whether or not rice serves as a status good or to achieve food security in the
case of own-production short-falls.

[i] Unconstrained Cereals Choices. Table 5-13 shows the relative proportions of
rice, millet, sorghum and maize household heads would on average have chosen "if they
had not produced any cereals of their own, and had a choice of 10 sacks which were to

last them until the next year’s harvest".l‘7 With an average of 3.3 sacks, rice ranked as

 

17'I'lne requirement that the 10 sacks selected had to last for one year was added to
incorporate storage problems with individual cereals. When farmers were subsequently
asked about reasons for their choices, storability did not appear to have been a major
consideration.

175

TABLE 5-13: UNCONSTRAINED CHOICE OF CEREALS FOR CONSUMPTION
(Standard Error)

 

Region Holof Fulani Total

 

 

North South No Yes No Yes

 

----------- Number of 50 kg Sacks----------

Rice 3. 4 3. 3 3. 3 3. 6 3. 6 3.2 3.3
( 2) ( 2) ( 2) (.3) (.2) (.2) (.1)
Millet 3. o 2. 4 2.5 3.6 3.2 2.3 2.7
( 2) ( 2) (.1) (.4) (.2) (-2) (.1)
Maize 2.1 2. 3 2.3 1.6 2.0 2.3 2.2
L 2) ( 2) (.1) (.3) (.2) (.1) (.1)
Sorghum 1. 5 z. o 1.9 1.1 1.3 2.1 1.8
(.2) ( l) (.1) (.3) (.2) (.2) (.1)

 

Soorog: ISRA/MSU Food Security Project Surveys (1986/87)

the preferred cereal, while maize was preferred over sorghum in both areas. Requiring
higher precipitation than millet, sorghum is gown mostly in the southern areas, and this
tends to be reflected in the choices.18 Wolof households (mostly in the north) also
appear to prefer millet over sorghum, while Fulani households prefer sorghum, relative
to members of other ethnic goups. Overall, maize does not appear to be perceived as
an inferior consumption item, and it is important to note that there is a strong prefe-
rence for a Enjoy of cereals.

Households heads were also asked why they had made each particular choice.
Among those selecting rice, roughly 50% responded that rice required less (or no)
pounding labor relative to coarse gains and that it was preferred for meals taken at

noon, visitors and special occasions. Non-rice cereals are usually threshed prior to

 

18Sorghum is viewed as somewhat inferior in the northern areas. Apparently the
historical role of this cereal in these areas (where sorghum-and rice-were gown when
rains were more plentiful) was to provide food to migant farm workers from Mali.

176
placement in sacks; had they been offered to farmers in unthreshed form, the propor-
tion favoring rice may have been higher. In contrast, households not choosing rice
indicated that it required expensive supplementary ingedients such as oil and other
condiments.
Millet and sorghum were preferred since it was perceived that they "remain in
the stomach” for a longer period of time, and were therefore particularly suitable during

the agicultural season.19

Maize was preferred since it yielded more flour than other
cereals, and because it tasted good when mixed with other cereals. Household heads
were also asked whether they preferred whole or broken rice; red or white sorghum and
yellow or white maize (Table 5—14). The first distinction tends to reflect preferences

between local and imported rice,20

while the latter are important from a production
perspective: the physical characteristics of red sorghum make it less desirable for birds
(it gows on husks with spikes which thwart bird attacks; and it tends to taste bitter
owing to a high tannin content). White maize is harder and more resistant to insect
attacks (and also provides higher yields) according to SODEFITEX extension agents and
farnners interviewed informally in the areas. However, yellow maize tastes better than
the more bland white maize, according to some of the farmers interviewed informally.
All (100%) of the Wolof households preferred broken rice, while preferences of non-
Wolof households were about equally divided between broken and whole rice. This may
reflect the fact that Wolof households no longer gow rice, and have developed a

preference for broken rice imported from Asia. Taken together, the results suggest that

 

19'Ilne digestibility problem of millet (i.e., that it takes longer to digest than maize or
rice), which reduces its consumption at noon in urban areas, therefore poses an advantage
for manual workers in rural areas.

20At least in urban areas, Thai broken rice is preferred since the increased surface area
more readily absorbs the sauce in the traditional Wong; dish.

177

TABLE 5-14: PREFERENCES AMONG CEREALS
Percent of Households

 

 

 

Region Total
North South N %
W
Hhole 14 67 95 45
Broken 84 32 113 54
Indiff. 2 1 3 1
Red 3 15 21 10
White 97 85 189 90
Indiff. l 1 O
Maize
Yellow 15 20 38 18
Hhite 78 77 163 77
Indiff. 7 3 10 5

 

Soozoo: ISRA/MSU FSP Surveys (1986/87)

rural households prefer variety in their consumption of cereals. So long as markets for
food remain unreliable, this may pose problems for a policy designed to induce farmers
to gow only a particular crop, such as maize.

[ii] Processing, Firewood and Storage Issues. Most of the northern (89%) and
three-quarters of the southern household heads reported that higher processing and
firewood demands of maize preparation reduced their overall consumption of this cereal.
Only 7% of the household heads—mostly in the north—had used a mechanical thresher to
shell and process maize at least once during 1986/87. In contrast, 27% of the northern
and 16% of the southern household heads had used a thresher to transform millet or
sorghum. Slightly more than one-half of the northern and 85% of the southern respon-
dents reported problems of storing maize (insects, rodents and humidity).

Two-thirds of the southern household heads also indicated that the task of

processing paddy rice into an edible form prevented them from consuming more local

178

rice, which is mostly gown by farmers themselves.21 Only 1 out of 124 southern
household heads had used a thresher to process rice (provided by SODAGRI at
Anambé). Southern household heads also reported storage problems with rice similar to
those of maize. In the northern areas rice is not stored for extended periods. Higher
processing and cooking requirements therefore tend to place maize at a disadvantage
relative to rice and millet or sorghum. This problem warrants further attention by policy
makers if maize production and consumption are to be expanded under the NAP.

5.1.5. The Special Role of Rice

Fifty-nine percent of the household heads reportedly increased, while 33%
decreased their rice consumption in 1986/ 87 over 1985 / 86. Among those reporting
increases in consumption, about one-half had produced more rice in 1986 (in the south),
while the remainder had increased their purchases of rice from the Gambia. In house-
holds with reduced rice consumption, 12% reported a decline in their own production of
rice, 8% reported a scarcity of rice in local markets, and 80% (mostly in the north)
replied they had produced more coarse gains in 1986, thereby reducing the need to pur-
chase rice. Since farmers were presumably still coping with the effects of the 1984
drought in 1985, this suggests that for some households Gambian rice was an important
way of assuring household survival. The finding also indicates that in rural areas, in
contrast to urban areas, coarse gains substitute substantially for rice.

To assess the independent effects of prices and income on rice purchases, a
reduced form regession model was estimated. The specification of the reduced form
and the results of the estimation are presented and discussed in Section 5.2.5. below

along with the results of an estimated marketed surplus function for coarse gains.

 

21Virtually none of the rice gown in the research areas entered cash markets.

179

5.1.6. Opinions of Market Reform

As the field work progessed, it became clear that the implementation of policy
reforms which this research project was to monitor was lagging belnind original time-
tables.22 Since this ruled out pre- and post-NAP comparisons of household behavior,
opirnion surveys were combined with surveys of potential behavior to anticipate potential
problems with the NAP. The results, presented in this section, are also useful as a check
on the reliability of the manifest data reported earlier.

To set the stage for this section, it is appropriate to consider the following
opinions expressed by farm household heads:23 ninety-nine percent ageed with each of
the following statements:

a. household heads should produce sufficient cereals to feed other

household members from one harvest period to the next;

b. it is better to gow one’s own cereals to feed the household because it

is diffith to save proceeds from sales of cash crops until the next year’s

hungy season; and

c. it is better to gow one’s own cereals, because even if one had the

financial means, one would not necessarily find cereals for consumption

on the market.

For a farmer who believes all other producers think alike, a general corollary of these
results is that it is unwise to produce cereals for sale in his or her area (see also the

discussion in Section 4.1.5.).

 

22m the southern areas the CSA failed to enter into action (and support producer
prices) in rural markets due to a lack of marketed surpluses.

23Questions were phrased as follows: ”Some farmers in the area say that Do you
agee with this declaration”?

180
[i] Opinions of Select Cereals Policy Measures. Many household heads,
particularly in the southern areas, were unaware of the producer floor price of 70 F / kg
for cereals (Table 5-15). Responses tend to reflect whether or not the CSA actually

operated in the area where the respondent lived.24

TABLE 5-15: KNOWLEDGE OF THE CEREALS FLOOR PRICE
(Three Years After its Announcement)
Percent of Households

 

Knowledge of
70 FCFA/kg North South Total
Floor Price

 

Unknown 20% 76% 109 52%
Known 77% 22% 94 45%
Incorrect 3% 2% 5 2%

 

Soonoo: ISRA/MSU FSP Surveys, 1986/7

Many farmers responded they would not change the area sown to cereals if the
official price of cereals were raised to 100 F/kg, lowered to 50 F/kg, or if the official
price were eliminated (Table 5-16).25

Some farmers reporting they would not respond to a price increase om have
been thinking of intensifying cereals production; among those who would increase cereals
area under any price policy scenario, many added that they attempt to increase the area
cultivated to cereals each year, regardless of prices. A few (northern) farmers felt the
floor price keeps prices artificially low.

For a variety of reasons, three-quarters (158/208) of the farm household heads

feel the government ought to assure a minimum cereals price to farmers. Some favor

 

24The CSA operated only in northern markets during 1987.

258cc Goetz, April 1988, for potential responses to peanut price changes.

181

TABLE 5-16: HON HOUSEHOLD HEADS SAY THEY HOULD
HOULD CHANGE AREAS SEEDED TO CEREALS UNDER
DIFFERENT OFFICIAL PRICE POLICIES
Percent of Households

 

 

 

 

 

 

 

Alternative Policies North South Total
100 F/kg: seed more 74% 64% 142 68%
seed less 0% 0% 0 0%
50 F/kg: seed more 9% 16% 27 13%
seed less 22% 2% 21 10%

 

 

 

Eliminate seed more 25% 30%. 58 28%
Floor Price: seed less «2% 2 1%

Soogoo: ISRA/MSU FSP Surveys (1986/87)

 

this policy believing it will facilitate buying food by making cereals more available
(Table 5-17).

Farmers in the southern areas on average felt the official minimum price should
be higher than did their counterparts in the northern areas (Table 5-18). This may
reflect their experience with historically higher cereal price levels in the area. In
comparison, Appendix A-7 shows farmers’ reservations prices for selling cereals out of

storage during July 1987.

TABLE 5-18: PREFERRED FLOOR PRICE LEVEL
BY AREA SEEDED TO CEREALS IN 1987, IN FCFA/KB
D-Deficit, SS-Self—sufficient, S-Surplus

North __.__59210_____
0 SS 5 0 SS S Total

Mean FCFA 63 75 81 86 87a 91 33
Std. Dev. 14 18 17 36 26 14 28

Valid N 6 24 39 72 29 6 176
Source: ISRA/MSU FSP Surveys (1986/87)

a. Data exclude 1 outlier of 500 FCFA/kg and are
for farmers willing to state a price (85%).

 

 

 

182

TABLE 5-17: REASONS GIVEN FOR AND AGAINST A CEREALS FLOOR PRICE
Percent of Households Reporting

 

 

 

 

 

 

 

 

 

 

 

 

Region

A. Favor Floor Price N S Total %
1. Provides a buying/selling reference point.. 13 19 32 26%
2. All products sold on the market should

have an official price ..................... 5 14 19 15%
3. Gives us more power relative to traders.... 13 5 18 15%
4. Induces producers to grow many cereals ..... 5 10 15 12%
5. Facilitates cereals trade ................. 14 O 14 11%
6. Helps everyone to have cereals ............. 3 6 9 7%
7. Helps us plan our production ............... O 5 5 4%
8. Stabilizes cereals trade ................... 4 0 4 3%
9. Cereals are our basic staple and must not

be sold at a high price .................... 1 2 3 2%
10. Provides us with cash income ............... 1 l 2 2%
11. Helps us buy a large quantity of cereals... 1 l 2 2%

Total valid responses: 60 63 123 100
Region

B. Disfavor Floor Price N S Total %
1. Favor free trade in cereals (without

public price policy) ....................... 3 10 13 39%
2. Producers should fix prices themselves ..... 1 9 10 30%
3. The price will be too low .................. 8 O 8 24%
4. The price will be too high ................. 2 O 2 6%

Total valid responses: 14 19 33 100

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)

Note:

Based on Open-Ended Questions

open-ended question.

Only a total of 156 valid responses were received to this

183
Household heads living in satellite villages were generally divided on whether or
not the distance to the nearest cereals market outlet prevented them from marketing
more cereals (Table 5-19). There are differences by triad, with all producers in the

Diaobe area responding that market access was a constraint.

TABLE 5-19: MARKET ACCESS AS A CONSTRAINT TO
SELLING MORE CEREALS

 

 

 

Poor Good Total

Access Access Percent
Research —
Triad No Yes No Yes No Yes

--# of Households-
Ndiapto 2 9 5 5 33% 67%
Ndoga 8 5 7 4 62% 38%
Diaobe O 11 O 7 0% 100%
Sandio 6 5 7 1 68% 32%
Diega 10 l 10 O 95% 5%

 

Source: ISRA/MSU Food Security Project
Surveys (1986/87)

Many respondents in the northern area favor the policy of free cereals trade
(Table 5-20). Table 5-21 gives reasons for this preference. As in the case of a cereals
floor price, some household heads favor this policy because they believe it would facili-

tate buying cereals.

TABLE 5-20: DESIRABILITY OF FREE CEREALS TRADE

 

 

Total
Opinion North South N %
% of Households
Disfavor 18% 53% 79 38%
Favor 67% 43% 110 53%
Don’t care 15% 4% 18 9%

 

Source: ISRA/MSU FSP Surveys (1986/87)

 

184

TABLE 5-21: REASONS FOR AND AGAINST FREE CEREALS TRADE

Percent of Households Reporting

 

 

 

 

 

 

 

 

 

 

Region
A. Favor Free Cereals Trade N S Total %
1. All areas will be supplied with cereals.... 20 2 22 25%
2. One can profit from cereals sales .......... 17 2 19 22%
3. It will be easy to get cash for cereals.... 10 6 16 18%
4. Can lead to price reductions among farmers. 0 10 10 11%
5. Hill allow us to seek best buy/sell prices. 5 1 6 7%
6. There will be no scarcity of cereals ....... 3 0 3 3%
7. Traders are more collaborative than Gvnt... O 3 3 3%
8. Hill increase the number of buy/sell places 0 3 3 3%
9. Producers will grow more cereals ........... O 2 2 2%
10. Traders can buy and resell during soodure.. O 2 2 2%
11. We were "fed up" with trading restrictions. 1 O 1 1%
Total valid responses: 56 31 87 100
Region

8. Do Not Favor Free Trade in Cereals N S Total %
1. Free cereals trade will lead to scarcity... 4 33 37 58%

2. Traders will transport cereals elsewhere
and/or resell dearly during the soudure.... 5 3 8 13%
3. Traders will speculate on prices ........... O 6 6 9%
4. It will not be profitable for us ........... 3 2 5 8%

5. There will be no more producers if every-
one becomes a cereals trader ............... O 3 3 5%

6. Producers will be spoiled since they will
speculate with cereals prices .............. O 2 2 3%
7. Banabaoo will not buy at official price.... 1 0 1 2%
8. Purchase price will be too high ............ l 0 1 2%
9. Producers will suffer if drought returns... 1 O 1 2%
Total valid responses: 15 49 64 100

 

Source: ISRA/MSU Food Security Project Surveys (1986/87)
Based on Open-ended Questions

185

The opinion survey results tend to complement the production and marketing
data reported earlier. Many households rely on coarse gain markets to purchase food,
even though these markets are perceived as unreliable. Given that these households
already face relatively high prices (above the floor price), the imposition of a floor price
is unlikely to improve the welfare of the set of farmers unable to produce enough food
under current prices. The remainder of this section examines household head opirnions
on how input/ output markets should be organized, and what they would do if they could
double their coarse gain production under the NAP.

[ii] Private Traders and Farmer Organizations: Which Mix? Many household
heads (87%) indicated it was a bad idea for the government-«i.e., parastatals-40 with-
draw from fertilizer distribution, and most (67%) did not believe private cereal traders

26 While the latter perception

would be able to take on this role in the short run.
presumably refers to physical distribution, it may be linked to the belief that traders
would be less forgiving than the government if cr0ps fail (given historical experiences),
and that they would charge higher interest rates for credit.

A further complication is introduced by the fact that 23% of the household heads
sampled believe farmers do not have a moral obligation to repay fertilizer credit in the
"event of crop failure. The question then is, will this belief change if private traders
rather than parastatals (commonly viewed as le Gouvomemont) deliver the input? If
not, traders will have problems when forced to repay their loans to commercial banks. It
is important to anticipate and provide rules covering these situations ex ante to avoid the

mistakes of past credit progams (i.e., unanticipated forgiving of debts). Whatever the

risk sharing rule, all progam participants need to be aware of it.

 

26Findings are consistent with results reported in Crawford (1987, et al., pg. 83) for
other areas of Senegal. The above questions were posed prior to the administration of the
APS-related questionnaire.

186
Most farmers favor receiving fertilizer and improved seeds distributed under the
APS through farmer organizations rather than directly from traders (Table 5-22),

regardless of whether the inputs are distributed on credit or on a cash basis.

TABLE 5-22: PREFERRED SOURCE OF INPUTS UNDER THE APS
Percent and Number of Households

 

 

Agent North South Total
Traders 6% 8% 15 7%
Farmer Org. 77% 90% 177 85%
Indifferent 17% 2% 17 8%

 

Source: ISRA/MSU Food Security Project Surveys
(1986/87)

Primary reasons given for this choice were that the solidarity of an organization
would lead to more equitable input distribution, more equal bargaining power (better
prices), and facilitate delivery of the inputs. Those in favor of receiving inputs directly
from traders cited a general distrust of farmer goups.

Various existing farmer organizations could, in the opinion of household heads,
assure the distribution of improved inputs to farmers. W was
cited most frequently in the northern areas (30%) while I’Associatioo Villageoise
received most votes in the southern area (51%) among those who felt existing farmer
organizations could take on this role (77% in the North, 99% in the South). More than
half (56%) the respondents in the northern area would not entrust the (peanut) gocto'oo
villageoise with the distribution, referring to problems in the past. In the southern area,
this proportion was only 22%.

To market their cereals, on the other hand, most farmers would prefer to by-
pass organizations and sell their output directly to traders (Table 5-23). Common

reasons were that farmers "had more trust in themselves" and preferred the flexibility of

187

TABLE 5-23: PREFERENCES FOR ORGANIZING SALES OF CEREALS
Percent and Number of Households Responding

 

 

Agent North South Total

Myself 60%] 57% 121 58%
Organization 36% 43% 84 40%
No Preference 5% 4 2%

 

Source: ISRA/MSU FSP Surveys (1986/87)

selling their output when, where, to whom and at prices they thought appropriate. Those
favoring sales through an organization indicated its credibility would thus be advanced; it
would be better equipped for dealing with traders; if one took credit from the organiza-
tion one should resell through it; and surplus organizational funds could serve those in
need (the latter response was cited mainly in the southern area).

The data thus suggest private food system participants would prefer to deal with
each other in an organized manner. Surveys of existing farmer organizations reveal,

27

however, most are ill-equipped to take on such tasks effectively. This in large part

explains why traders would prefer to organize farmers themselves, rather than rely on
existing goups.28
[iii] Marketing Responses Under Doubled Cereals Production. This section
presents data on the proportions of cereals output farmers claim they would sell if they
could double their production; the timing of those sales; and the uses of proceeds from

the sales. Household heads were asked, if they could double the amount of cereals

produced in 1987 under the APS, which proportion of the total production would they

 

27A notable exception are the ABP’s organized by SODEFITEX, but drawing on them
may lead to "turf battles". See Diagana (June 1988) for survey results pertaining to existing
farmer organizations.

28See Goetz et al. (September 1988) for private traders’ opinions of the NAP and how
they would prefer to organize input/output markets.

188
anticipate selling. For all farmers, including those not willing to sell any particular cereal
(as well as those not willing to buy the inputs at currently proposed prices, cf. Ch. III),
average percentages of total production they would sell at 70 FCFA/ kg range from 26 to
30% (depending on the cereal; minimum=0%, maximums-100%). This corresponds to
52 - 60% of 1986 production, representing a sizeable increase in quantities marketed (cf.
Table 5-1).

As indicated earlier, not all farmers will adopt the improved inputs and be able
to increase their cereals production. Nevertheless, even small changes in coarse gain
sales will depress prices if markets do not become better integated over time and space,
making it costly for the CSA to maintain a floor price (as the experience in the market
of Ndoga Babacar, Maka in 1986/87 shows). In turn, lower prices received by farmers
will alter the goss benefit cost ratios for using fertilizer presented earlier (Ch. III),
raising questions about the profitability of fertilizer use.

Most farmers responded they would not sell their cereals at harvest, which could
reinforce the sharp seasonal price decline observed in 1986/87. Instead, the following
periods were listed, each with similar frequencies: 1. cash crop marketing season
(presumably when food-deficit households would be able to buy cereals from proceeds of
cash crop sales); 2. after that period but before the beginning of the rainy season; 3. at
the beginning of the rainy season; and 4. during the gem (hungy season), when
prices traditionally reach their seasonal peaks. Table 5-24 suggests purchases of livestock
would constitute the most important use of funds in the post-harvest period.

The above analysis is partial, and based largely on potential behavior. Neverthe-
less, a picture of the complex economic system in southeastern Senegal emerges. When
one sector in this system is perturbed consequences will be felt elsewhere, such as in the
off-farm activity and livestock sectors. Prudent development planning requires at least

minimal knowledge of all of these sectors to anticipate consequences-good and bad--of

189

TABLE 5-24: PRIORITY USES OF PROCEEDS FROM SALES OF SURPLUS CEREALS
(Post-Harvest Period)
Number and Percent of Households

 

 

 

 

 

 

 

 

North South Total

First Use

Purchase Livestock ......... 33 43% 34 34% 67 38%
Pay Labor/Debts ............ 4 5% 23 23% 27 15%
Buy Equipment/Animals ...... 13 17% 6 6% 19 11%
Other Non-Agric. Use ....... 5 6% 10 10% 15 8%
Save for Need/Medicine ..... O 0% 14 14% 14 8%
Buy Clothes ................ 9 12% 3 3% 12 7%
Buy Fertilizer ............. 4 5% 3 3% 7 4%
Construct House/Marry ...... 6 8% 3 3% 9 5%
Other Agric. Use ........... 3 4% 5 5% 8 4%
Total................. ..... 77 100% 101 100% 178 100%
Second Use

Purchase Livestock ......... 18 26% 16 20% 34 22%
Buy Equipment/Animals ...... 17 24% 11 13% 28 18%
Other Non-Agric. Use ....... ll 16% 12 15% 23 15%
Save for Need/Medicine ..... 2 3% 15 18% 17 11%
Buy Clothes ................ 6 9% 6 7% 12 8%
Construct House/Marry ...... 8 11% 4 5% 12 8%
Pay Labor/Debts ............ 2 3% 8 10% 10 7%
Buy Fertilizer ............. 2 3% 5 6% 7 5%
Other Agric. Use ........... 4 5% 5 5% 9 5%
Total...................... 70 100% 82 100% 152 100%
Third Use

Buy Equipment/Animals ...... 21 38% 11 20% 32 29%
Other Non-Agric. Use ....... 8 15% 12 22% 20 18%
Save for Need/Medicine ..... 8 15% 7 13% 15 14%
Purchase Livestock ......... 4 7% 8 15% 12 11%
Construct House/Marry ...... 7 13% 3 6% 10 9%
Buy Fertilizer ............. 3 5% 5 9% 8 7%
Pay Labor/Debts ............ 1 2% 4 7% 5 5%
Other Agric. Use ........... O 0% 4 7% 4 4%
Buy Clothes ................ 3 5% 0 0% 3 3%
Total... ................... 55 100% 54 100% 109 100%

 

Sourco: ISRA/MSU FSP Surveys (1986/87)

190
individual progams and to identify strategic points of leverage in the system where
progam dollars will have a high pay-off.

To address the criticism that the analysis thus far is partial, the independent
effects of various explanatory variables on households’ coarse gain marketing behavior
is now examined. In particular, we are interested in understanding the households’ net
position in the market, and in examining the implications of this phenomenon.

5.2. MARKETING EQUATIONS FOR COARSE GRAINS AND RICE

Aside from complicating the implementation of floor price policies, the tricho-
tomy of households-met buyers, net sellers, non-participants--also raises questions for
econometric modelling: which selection mechanism(s) determines whether or not a
household participates in coarse gain cash markets; do net buyers and net sellers
respond the same way to exogenous factors (e.g., prices), and is the continuous decision
of how much to sell or buy and the discrete decision of whether or not to participate in
cash markets influenced by the same set of variables? Implicit in most of the empirical
literature on marketed surpluses of food in Africa is the assumption that observations
are drawn from a homogenous population in which agents respond in identical fashion,
for example, to a change in relative prices.

5.2.1. Recent Estimation ElTorts

A recent attempt to estimate the marketed surpluses of agicultural households
in West Africa is found in Strauss (1984). Building on earlier work by Krishna (1962);
Lau, Lin and Yotopoulos (1978); and Barnum and Squire (1979), Strauss postulates the
model

(1) xi X,(p.z.k)

(2) X,“ xflpm + p,T(m) + f(pz.k)l

where Xi refers to production and Xic to consumption of good i, p is a vector of prices,

pn is the price of labor, n denotes household characteristics affecting taste, T is time

191
available to the household for work and leisure, m is household characteristics determi-
ning T, 2 is a vector of farm characteristics including fixed inputs, k is a vector of
production technology parameters, A is exogenous income, and f is profits. From (1)
and (2) the absolute value of the marketed surplus of commodity i is calculated as
(3) IMsil = X,(-) - xi°(-)

This yields the marketed surplus elasticity of good i with respect to the price of good j,

 

 

(. _. __Xi 3.1.5."; _ 6° .21 ”6°
IMSiI 6pj IMSiI Xi6pj lMsifilx 6p]-

which represents the difference between household supply (production) and demand
(consumption) elasticities, weighted respectively by the ratios of quantities supplied and
demanded to the absolute value of the marketed surplus. Note that for households not
participating in the market for commodity i the elasticity is undefined.29

Implicitly assuming that production and consumption decisions are made simul-
taneously (see also Singh et al., 1986), Strauss derives the reduced form of the marketed
surplus as

(5) MSi = MSi(p,n,m,z,k,A).
As an alternative, Strauss proposes directly estimating the underlying structural supply
(Xi ) and demand (X ic) equations. One advantage of doing so is that the deriveduor
restricted--reduced form estimator of II, i..,e I! = - 2 I’ll, where A and 1"1 are consistent
estimates of the structural parameters in the system YI‘ + XA + e = 0, is asymptotically

efficient relative to the 01.8 reduced form estimate, 1! = (X’X)'1X’Y when some

structural equations are overidentified (see Schmidt, p.241 for a proof; Junankar

 

29As Strauss points out (p. 326), however, in a latent variable model where Yi
max(0, Y i") the unconditional expectation E[Y] = E[Y IYi -->0] P(Yi >0). McDonald and
Moffit discuss the interpretation and use of Tinbit coe 1cients.

192
discusses problems of estimating profit functions given data limitations and market
imperfections in developing countries). Marketed surplus elasticities are then calculated
using equation (4), allowing price changes to affect both consumption and production
decisions through profit effects.

The interdependent, albeit sequential, determination of production and consump-
tion decisions in Strauss’ formulation, and the separability of these functions for estima-
tion purposes, assumes that there is no uncertainty, that there are markets for all inputs
and outputs, and that the household may participate with minimal cost in these markets.
For example, with perfect knowledge of yields and market prices in the post-harvest
season, a household may decide at planting to produce only one-half of its food needs,
and to generate sufficient cash crop income to in turn make up the food production
deficit through purchases in the post-harvest season.

5.2.2. Model Specification Under Uncertainty

Given uncertain rainfall, thin and often non-existent food markets and severe
credit constraints in subsaharan Africa, the assumptions made in Strauss’ model are
strong. Observational evidence and opinion surveys of household heads in Senegal and
Mali (see Dioné, 1989) reveal that coarse gain markets are perceived as unreliable both
as a source of and an outlet for food.30 This suggests that households in Senegal strive
for at least some degee of self-sufficiency, and that own-production levels do influence
the decision to consume and sell or buy coarse gains, once crops have been harvested.
Hammer develops an intertemporal model of household food self-sufficiency behavior
under yield uncertainty, and (successfully) tests the model using time series data from

Senegal. Shaffer et al. (1983) and Binswanger et al. (1987) attempt to explain the

 

30In one market shed studied, for example, coarse gain prices averaged 100 FCFA/ kg
in August 1986, and 45 FCFA/ kg in August 1987. Staatz et al. (1989) report similar survey
findings in Mali.

193

unreliability of food markets and argue this phenomenon is widespread in land-abundant
subsaharan Africa. Uncertainty is thus pervasive and has to be dealt with analytically.

The basic reduced form equation (5) proposed by Strauss is estimated here, but
own-production levels rather than technology parameters are included as predetermined
variables. Variables affecting the household’s (agicultural) labor-leisure choice are
excluded under the assumption that this problem was solved during the preceding
gowing season. Marketed surplus (MSi) is then specified as a function both of 1) price
expectations which, for lack of a better measure are equated with average market prices
observed in the post-harvest season, i.e., Et_1[pt] = B-pt, where 8!] and Et-l denotes
expectations at planting about post-harvest season prices; 2) entitlements E which
include own-production levels of crops but also obligations incurred during the gowing

88350“. 81161! as borrowing of draft equipment and cash;31

and 3) consumption preferen-
ces and other household characteristics affecting marketed surpluses of coarse gains.

This leads to a specification in which behavior during the marketing season
depends on relative prices p, including expectations at planting although they are not
distinguished from observed prices, current period constraints B such as own-production
and household characteristics ll which include the age of the household head, the
number of consumers per active worker and ethnicity,

(6) MSi = MSi(p,E,I'I) + E

The distribution of the error term 6 is discussed in Section 5.2.3.

Market prices, however, are neither uni-dimensional nor costless to observe for

individual households. Given genuine uncertainty, costs of observing market outcomes,

 

31h is not known whether these transactions, in the terminology of Bronnley and Chavas
(189, p.721), are contingent on harvest outcomes or unconditional such that the terms of
exchange are known when the transactions is made. If these borrowing decisions are made
simultaneously with subsequent coarse gain marketing decisions, a simultaneous equation
bias arises.

194

and following the argument in Heiner (1983), the question of how much a household
sells or buys is cast as: When is it the right time to increase or decrease stocks held?32

If this is a more adequate formulation of the question, it appears that Strauss is
only partly correct in criticizing Bardhan (1970) and Haessel (1975) for including
quantities produced in their equations: "[the authors] assumed that production was fixed;
hence 20W" (emphasis added)-
Household food storage is an empirical regularity in subsaharan Africa, and sales or
purchases can be viewed as depleting or adding to these stocks33 (see Renkow, 1988, for
a similar analysis recognizing that households manage inventories of food and the
discussion with regard to price uncertainty in Singh et al., p.57).34 Hence price changes
will lead not only to relative changes in the consumption of individual commodities, but
also to changes in stocks. A higher price of food may lead to increased sales of food,
due to a higher opportunity cost of holding stocks, but it may also lead to purchases of
food if it signals impending scarcity. Similarly, the availability of coarse gains proces-
sing technology may increase the opportunity cost of holding unthreshed stocks relative
to the cost of threshing coarse gains, and thus also represents a change in the relative
price of coarse gains to that of rice (for example).

Another dimension of observed prices is market access. Poor market access due
to lack of transport and/ or distance not only reduces [increases] the net (of transport
cost) price received [paid] by the seller [buyer], thus lowering [raising] the opportunity

cost of holding stocks, but also increases the farmer’s cost of observing market prices in

 

32’I'herefore, rather than treating marketed surplus solely as a "static" planned quantity,
it is viewed here as a flow from a buffer stock to cope with uncertainty.

33Strauss assumed beginning and ending stocks were zero. In this study ending stocks
were found to be positive for most households. Borrowing of cash (mostly for consumption
purposes) during the preceding gowing season is used as a proxy for low beginning stocks.

34Binswanger et al. (1987) argue that stocks of food and livestock holdings are major
forms of capital accumulation in land-abundant agiculture.

195
order to make transaction decisions. Conversely, a household pursuing off-farm activi-
ties may find it less costly to observe prices and therefore be more likely to participate
in markets, ootou's DAM-

Consequently, there are two major differences between the estimation attempted
here and that pursued in the agicultural household literature (which is based on systems
of demand and supply equations). The first variation from previous work is explicit
allowance for genuine uncertainty (i.e., that which cannot be parameterized), which leads
to a number of new variables as regessors. The second variation results from transac-
tions costs. It builds on the uncertainty dimension, and is discussed in the following
section.

5.2.3. Model Specification with Transactions Costs

The non-market participation of households discussed earlier is largely explained
by transactions costs. These costs drive a (shaded) wedge between household buying and
selling prices, as shown in Figure 5-4, which is based on the concept of non-traded goods
in the international trade literature (Dornbusch, 1980).35 The y-axis shows the market
price of food paid or received by a household participating in the cash market. The x-
axis shows the value (price) of food to the household. It there are zero transaction costs
(1 = O), the household can equate its shadow price with the market price, and market
participation behavior is continuous (as opposed to discrete or subject to a threshold) as
both prices vary.

For any transaction cost 1 > 0, however, we expect to observe a non-empty
autarky set {A} e ¢. In other words, there will be a set of households which cannot

equate market and shadow prices (over a certain range) owing to transaction costs. Only

 

35A similar analysis examining the behavior of households in the market for land
tenancy is given in Bell et al. (1988).

196

FIGURE 5-4: SELLING, PURCHASING AND AUTARKIC BEHAVIOR OF HOUSEHOLDS
AS A FUNCTION OF MARKET AND SHADON PRICES AND TRANSACTIONS COSTS ‘r

 

 

 

 

 

 

Market
Price Pm . ,.
'fi-Ps-Pm(l+1)
P o
m '5'.) I
'15/ 1
..::./..._‘. . :
..,..U I
:4 I 1
-/ I I
/ l I
1 1 Shadow
0 E P Price Ps

Note: transactions costs (7) are assumed to be a fixed
percentage of volumes transacted.
Sooooo: Based on Dornbusch (1980).

as 1’ -> O, {A} = d as households are able to equate the internal shadow price ps with
market price pm. As 1 -> 0, precluding exchange, the set {B} =- {S} = 0.

Transactions costs therefore lead to an important econometric question: how is
the distribution of the error term 6 to be specified in light of the sample separation?
Do we assume that observations are drawn from identical distributions so that buyers,
for example, exhibit the same price elasticity as sellers, conditional on market
participation?

Figure 5-5 shows the implications of alternative assumptions about the transac-
tion cost parameter 1 for specifying the error term density. For ease of exposition it is
assumed that net marketed surpluses of buyers and sellers are iid ~ N( I MSslps) =

N( I MSbI,ob). Case 1 shows two symmetric densities for buyers and sellers for the

197

FIGURE 5-5: PROBABILITY DENSITY FUNCTIONS FOR QUANTITIES BOUGHT
AND SOLD UNDER ALTERNATIVE TRANSACTIONS COST ASSUMPTIONS

Note: Identical (symmetric) densities assumed for simplicity.

T-transactions cost; E-expectation operator; Q =quantity
sold; Opaquantity purchased; PS-shadow, Pm-marfiet price.

Case 1: t s O p

p----- ----

 

 

3

( ..

o ‘--------—

 

 

 

 

 

Purchases E[er Autarky EI : Sales
(P: 2?.) (Pa '9.) (Pa (Pa)
Case 2: O t t~< - p
a I
I 1
: I
g 3
. 1
1 I
1 1
1 I
t L‘. _ ' _ >
Purchases E[Qp] Autarky E[Qs] Sales
(P: 2?.) (Pa ‘9.) (Pa (Pa)
Case 3: lie t eI- p
< n
Purchases Autarky Sales
(Ps’Pul (Ps’ipal {P3(Pal

file=dis.fig E[Qp]=E[Qs]=O

198

hypothetical situation where 1 = 0 and the observations have been drawn from two
heterogeneous populations, allowing for different responses to the same exogenous
variable in the estimation of parameters. Here {A} =- ¢ since all households will trade.
In case 2 the densities will slide toward each other as 1 increases. The set {A} has
support for T > 0, since some households k will not participate in the market:

{A} = {klpmk/(1+Tk) < Psk < Pmk(1+7k)} G 2
We cannot determine whether households k c A are "closer” to set {B} or {S} although
one conjecture is that the higher (lower) food output per capita, the closer the household
will be to set {S} ({8}). Case 3, finally, shows the situation where 1' -> no, such that
expected quantities traded become zero.

The classical transactions costs or friction model, developed by Rosett (1959) to
explain why asset holders will not adjust portfolios in response to small changes in asset
yields, is specified as (using the notation in Maddala, 1988):

(7) yi“ = B’xi + "i

yi = yi" - al ifyi“ < 01 (i155)

(8) yi = 0 ifal s yi‘ s 012 (115A)

yi = yi“ -a2 ifaz < yi’ (ieB)
where yi" -- desired change in coarse gains (or asset) holdings of individual i; yi =
actual change in coarse gains holdings (sales and purchases); xi = exogenous variables;
[3 = coefficients to be estimated; ui = random error term, assumed to be ~ N(0,a'2); al
(<0) = desired decrease in coarse gains holdings; (12 (>0) = desired increase in coarse
gains holdings; ¢(-) = standard normal density function; and §(.) = cumulative

standard normal density function. The associated likelihood function is

(9) L(al.az.B.0|y,X) =
= H 3,60%," xi) 1‘1 [il—L—M ’3 X') - §(—1-——*“ ‘3 x‘1] 1'73,“in
ies ieA IEB

'B’Xi
a .)

 

 

199
The problem with this specification for the present purpose is that asset yield
variables Xi entering the likelihood function are assumed to be identical for each agent.

For example, intra~day changes in asset prices are observable at a low cost once the fixed

cost of commissioning a broker have been incurred. In subsaharan Africa it is common-

ly observed that market prices vary considerably within a market day; that they tend to
be agent-specific (see Scth and Robison, 1988, on arms-length transactions in the

U.S.); and that roundabout trades-e.g., livestock for peanuts for foodufurther obscure

actual prices paid and received by individuals. This means measured market prices are

indicative proxies for price levels prevailing in any individual transaction. These observa-
tions suggest the following may be a more fruitful approach to answering the questions
raised earlier, i.e., what determines whether or not a household participates in markets
and, if so, how much it buys or sells.

[a] A probit model designed to discern the probability of market participation under
the maintained hypothesis that net buyers and net sellers react in the same
manner to the effects of exogenous influences (although this too can be dealt
with in Rosett’s (1959) model):

(10) P(Y=1) = II>(B’X)
where (I is the cumulative standard normal distribution function, Y=1 if the
household buys or sells, and Y=0 otherwise.

[b] Two OLS equations, one for the entire sample and one only for those participa-
ting as net buyers or net sellers:

(11) Y = B’X + E
where Y denotes net sales of coarse gains and 6 ~ N(0,02).

[c] A trinomial logit equation with three possible (unordered and mutually exclusive)

outcomes reflecting the trichotomy of households: not in the market, net buyers

and net sellers.

[d]

[e]

[f]

200
(12) P(Y=m) = exp(B’xm) / [Zmexp(B’xm)], m = 0,1,2 with Bm=0.
Two tobit equations, one for net buyers and one for net sellers, using households
not participating in the coarse gain market as the zero category in both cases.

Here the likelihood functions are (note the similarity to Rosett’s model)

(13) L(B,.o,) - 1‘1 [1 - 6(B—'§I)] 1‘1 -¢(y—I——-4 ‘ 0'“ ")
keA keS

(14) L(B,.ab) = fl [1 - 61%)] 1‘1 EMF—461 ' ,3 " 1

keA keB
Along the lines of Cragg’s (1971) model (see Haines et al. (1988) for a recent
application to consumption behavior) two separate probit equations are esti-
mated predicting the probability of market participation for net sellers and net
buyers, and
two truncated equations where E is explicitly truncated at zero, to examine the
continuous behavior of households conditional on being net buyers or net sellers.
5.2.4. Empirical Results

Table 5-25 presents estimated coefficients for equations (10) and (11) under the

maintained hypothesis that households respond in the same manner to environmental

circumstances, regardless of whether they are net buyers or net sellers.36 Particularly

noteworthy is the result that households with off-farm incomes are morojkoly to

participate in the market, but that this factor tends not to affect the ooomjty of net sales

significantly. As hypothesized earlier, this may be due to reduced costs of observing

coarse gain prices and/or speculative activity, effects not picked up in the 01.8 equa-

tions which assumes continuous adjustments by farmers.

 

36Wages and livestock prices were omitted due to multicollinearity with rice prices.

Cash crop (cotton and peanut) prices are fixed pan-territorially by the government.

201

TABLE 5-25: PROBABILITY OF PARTICIPATING IN COARSE GRAINS
CASH MARKETS VERSUS NET QUANTITIES SOLD
(absolute t-statistics)

 

Net Buyers 8
Sellers Only

Dependent Variable= Entire Sample
Net Sales (0.1

and/or continuous)

 

Trinomial Logit

 

 

 

 

 

 

Exogenous Variable Probit OLS Buyers Sellers OLS
Constant 4.86 1136 1.33 15.6 1588
(2.46) (2.66) (0.324) (3.49) (2.83)
Coarse Grain Price 0.0356 6.60 0.0873 0.0650 7.95
(CFA/kg) (2.25) (2.05) (2.47) (1.60) (1.74)
Rice Price -0.0415 -10.1 -0.0430 -0.135 -13 2
(CFA/kg) (2.97) (3.76) (1.44) (3.94) (3.47)
Poor Market -0.223 -3.68 -0.513 -0.351 22.8
Access-l (0.675) (0.048) (0.833) (0.441) (0.227)
Mechanical Trans- 0.0820 192 -1.28 1.15 299
formation Avlb. (0.176) (1.86) (1.17) (1.12) (2.23)
Off-Farm Income=1 1.31 103 2.67 2.32 94.5
(2.90) (1.26) (2.66) (2.39) (0.829)
Fulani Ethnic -0.205 -178 -0.234 -0.684 -198
Group=1 (0.524) (2.20) (0.310) (0.841) (1.92)
Coarse Grain Output 0.0000555 0.0739 -0.00018 0.000456 0.857
(kgs) (0.470) (2.81) (0.652) (1.76) (2.42)
Cash Crop Output -0.00180 -0.396 -0.00333 -0.00035 -.433
('000 CFA) (2.11) (2.38) (1.47) (2.05) (1.92)
Consumers/Active -0.0852 -121 -0.0309 0.476 ~163
(0.205) (1.43) (0.038) (0.516) (1.39)
Household Head Age -0.00453 2.79 -0.00412 0.00420 1.83
(0.452) (1.23) (0.208) (0.189) (0.630)
Traction Equipment 0.643 201 0.250 3.33 273
Borrowing HHsI (1.46) (2.31) (0.302) (2.99) (2.48)
Borrowed '000 CFA in 0.00534 -2.16 0.0147 0.00698 -2.26
1986 Hungry Season (0.572) (3.61) (0.856) (0.403) (3.36)
Nbr. of Observations 116 116 116 85
Adjusted R-Square 0.35 0.43
Log-Likelihood 56 84
Prediction Successzpao 29% 42% 23%
p81 95% 80% 80%

 

202

Conversely, the availability of mechanical transformation, Fulani ethnicity, coarse
gains output and the borrowing of equipment and credit during the 1986 gowing season
affect net sales significantly but appear not to enter the market participation decision
(agairn, comparing the OLS and limited dependent variables models in Table 5-25). Cash
crop markets tend to be physically separate from, and of shorter duration than, coarse
gain markets. Hence increased participation in these markets does not convey informa-
tion on food prices. The probit equation performs poorly in predicting when households
do not participate in markets (prediction success of only 29%). Presumably this is due
to attempting to predict market participation for both buyers and sellers. The trinomial
logit specification suggests net buyers and sellers do react differently to exogenous
influences, although most of the coefficient estimates are insignificant at the 10% level.

Results for the case where net buyers and sellers are postulated to have been
drawn from different populations are reported in Table 5-26. For the not in the mar-
ket/ net sellers equation the tobit equation performs fairly well, except that it has
difficulty distinguishing discrete from continuous behavior for the availability of mechani-
cal transformation, off-farm income, cash crop production and borrowing during the
hungy season. For households not in the market/net buyers the tobit equation yields
dismal results, showing only one significant coefficient (borrowing during the 1986
gowing season). Of particular interest is the result that a higher coarse gains price
significantly increases the probability of market participation but, conditional on market
participation, a higher price reduces quantities purchased (note that these results have to
be interpreted with caution since market prices are based on only 5 observations or
market places). This may reflect the fact that a higher price signals impending scarcity
for these households, which on average had produced less coarse gains per capita
(Table 5-2), so that they entered the market as buyers. For both sets of households,

therefore, a higher coarse gain price increases the probability of market participation,

203

TABLE 5-26: DETERMINANTS OF HOUSEHOLD BEHAVIOR IN THE

COARSE GRAIN CASH MARKET

(absolute t-statistic)

 

Not In Market/Net Sellers

 

 

 

Exogenous Variable Tobit
Constant 1895
(3.33)
Coarse Grain Price 15.2
(CFA) (2.73)
Rice Price -18.5
(CFA) (4.70)
Poor Market ~47.7
Access-l (0.423)
Mechanical 276
Transformation Avlb. (2.29)
Off-Farm lncomesl 172
(1.70)
Fulani Ethnic -128
Groupsl (1.19)
Coarse Grain Output 0.0738
(kgs) (2.15)
Cash Crop Output -0.441
('000 CFA) (2.30)
Consumers/Active -182
(1.72)
Household Head Age -1.09
(0.365)
Traction Equipment 526
Borrowing HH-l (3.72)
Borrowed '000 CFA in 5.36
1986 Hungry Season (2.29)
Nbr. of Observations 71
Log-Likelihood 298

Prediction Successzpao

Probit

 

9.74
(3.49)

0.0523
(1.86)

-0.862
(3.77)

-0.0406
(0.774)

0.723
(1.13)

1.55
(2.32)

-0.664
(1.23)

.000274
(1.61)

-.00235
(2.03)

-0.427
(0.720)

-.00177
(0.123)

236
(309)

0.0150
(0.905)

 

71
27
74%
95%

Trunc

 

1973
(1.78)

49.7
(3.26)

-31.3
(3.04)

-119
(0.476)

434
(2.01)

~6.89
(0.034)

-81.4
(0.397)

0.0988
(1.54)

-0.338
(0.900)

-461
(2.06)

-7.82
(1.13)

988
(2.81)

10.5
(2.57)
40

258

Not In Market/Net Buyers

 

Tobit

 

-501
(0.579)

1.14
(0.213)

2.45
(0.428)

24.2
(0.202)

-96.1
(0.439)

167
(1.01)

116
(0.814)

-0.0436
(0.853)

-0.149
(0.358)

86.7
(0.565)

-3.78
(0.946)

-22.5
(0.158)

2.71
(3.43)
76

351

 

 

 

Probit Trunc
0.984 ~5717
(0.371) (2 31)
0.0419 -31.8
(2.48) (2.33)
-0.0216 49.5
(1.18) (2.77)
-0.169 114
(0.444) (0.557)
-0 392 356
(0.577) (0.755)
1.44 -724
(2.48) (1.74)
-O.273 -153
(0.555) (0.547)
-0.000044 0.0797
(0.266) (0.762)
-0.00243 1.72
(1.62) (1.64)
-0.0444 430
(0.088) (1.35)
-0.00267 -17.8
(0.220) (2.02)
0.175 23.1
(0.342) (0.081)
0.00506 3.29
(1.05) (2.58)
76 45
41 298

55%

84%

 

204
but sellers sell more while buyers buy less, as we might expect, once they have decided to
participate. The problem of the tobit model in predicting the effect of off-farm income
arises from the fact that off-farm activities increase the probability of participation while,
conditional on being in coarse gain markets, households with off-farm income buy less
than those without such income (significant at the 10% level).

The results thus suggest that off-farm activities increase the probability of market
participation by reducing the cost of observing prices, and imply some specialization
among households, with one goup entering markets as net sellers and the other as net
buyers, depending on the nature of the off-farm activity. In a more dynamic sense, off-
farm activities may also allow households to invest in production-increasing inputs, even
though in this data set off-farm income did not appear to be strongly associated with
equipment ownership or the use of mineral fertilizer (Ch. III). These hypotheses could
be tested in future research, and it would be important to follow a panel of households
over multiple years.

The tobit model also has difficulty discerning the effect of cash crop output
(which is highly correlated with cash crop sales, with a correlation coefficient” of 0.968)
in the case of coarse gains buyers. The more cash crops produced, the lower the
probability of participation, but once transactions costs have been incurred, households
buy more coarse gains for every 1000 FCFA of cash crops produced (and sold). For
the sample these coefficients are marginally significant at the 10% level.

For both buyers and sellers, having borrowed money (mostly for consumption
purposes and unexpected family needs) led to significantly higher sales and purchases,
conditional on market participation. It is conceivable that sellers lacked alternative

means of repaying loans, while buyers had such alternatives and instead stocked food to

 

37Signii'1cant at the .001 level.

205
avoid mistakes of the past hungy season (i.e., running out of food). This is an example
of households responding in opposite fashion (sell more, buy more) to the same exoge-
nous stimulus, suggesting that buyers and sellers were indeed drawn from different
populations.38 On the other hand, the amount of coarse gains produced tends to
significantly increase both the probability of market participation of sellers and quantities
sold (tobit coefficient), but plays no significant role in the case of buyers in this sample.

5.2.5. The Rice Regression Equation

In the rice regession equation livestock and draft equipment ownership are
introduced as proxies for wealth. Since 16% of the households did not purchase rice, a
tobit model was employed.

Results shown in Table 5-27 suggest rice prices did not significantly affect the
probability and level of rice purchases, while higher coarse gain prices reduced rice
purchases. The latter result is surprising and may reflect strong income effects between
rice and coarse gains. Using equipment value as a proxy for long run wealth suggests
wealthier households purchase significantly more rice, although this is not true for
livestock ownership as a proxy for wealth. Similarly, short run income, measured by cash
crop production levels, does not appear to affect rice purchases. The availability of
mechanical coarse gains processing technology and higher own-production levels signifi-
cantly reduce rice purchases. The first phenomenon is likely to reflect the reduced cost
of processing coarse gains into an edible form. Households headed by older farmers
bought significantly less rice, ceteris pagjous, conceivably reflecting weaker preferences
for broken rice.

It is interesting to note, finally, that the coefficient estimate for off-farm income

is not significantly different from zero in the rice equation, suggesting involvement in

 

38Survey results indicate the net market positions were fairly stable over the period
1985-87.

206

TABLE 5-27: REDUCED-FORM ESTIMATES FOR RICE PURCHASES
[Dependent Variable-kgs Rice purchased; (t-statistics)]

 

 

 

 

Exogenous Variable Tobit
Constant 494
(3.22)

Rice Price 0.644
(0.685)

Coarse Grain Price -4.63
(4.13)

Equipment Value 4.72
(4.39)

Livestock Owner 18.4
(0.753)

Fulani Household -7.93
(0.286)

Household Head Age -2.02
(2.51)

Consumer/Active -43.6
(1.55)

Off-Farm Income 6.43
(0.234)

Poor Market Access -88.1
(3.38)

Mech. Tran. Techn. -112
(3.09)

Coarse Grain Harvest -0.0231
(2.37)
Cash Crop Harvest 0.0280
(0.472)

Rice Harvest -0.0481
(0.703)

Log Likelihood 603

 

207

market activities does not provide additional information which reduces the cost of
market participation. This is plausible in the case of rice markets which, although often
clandestine, appear to convey information readily. Part of the proper functioning of the
market may result from the lowered uncertainty stemming from an almost perfectly
elastic supply of rice on world markets.

5.2.6. Policy Implications

The following implications may be drawn for policy makers in Senegal. Less
costly market information appears to increase the probability that a household will
participate in coarse gain markets, as reflected in the coefficient on the off-farm income
variable (which may also measure an increased willingness and ability of farmers to
manipulate their stocks of food through purchases and sales). To the extent that
exchange is beneficial, the social returns to increased market information have to be
weighed against the cost of providing such services in sparsely populated rural areas.

Higher coarse gain prices increase market participation but, as suggested above,
price signals are interpreted differently by buyers and sellers. It makes a difference
whether households perceive the price change to be caused by market forces (scarcity)
or as a temporary aberration due to government intervention in markets (these results
have to be viewed with caution owing to the limited price data). Promoting coarse gain
threshing technology in rural areas may stimulate sales of coarse gains, but according to
the significance of the t-statistics, would benefit mainly net sellers in the short run. This
is to be expected, since threshing is a supply-side taskuas opposed to a demand-side task
such as milling.

Higher cash crop sales reduce the probability of participation in coarse gains
markets as sellers, thereby detracting from the objective of increasing marketed surpluses
of food. However, because of (possibly sizeable) economies of scope in cash and food

crop production, and the peanut seed-labor linkage, we cannot recommend a reduced

208
emphasis on cash crop input delivery and production in the short run. Reducing trans-
actions costs in cash crop markets with regard to access, queuing, patronage and irregu-
larities at scales may free farmers’ time which could be better invested in ascertaining
coarse gains prices.
5.3. SUMMARY AND CONCLUSION

Results presented in this chapter suggest that households in southeastern Senegal
pursue a variety of strategies to achieve food security, and that over 30% rely on coarse
gain markets to purchase some food. It was argued that some of these households will
have difficulty responding to a floor price incentive, and that such a policy would have
only narrow distributional benefits. Opinion surveys regarding free trade policies
confirm this general result, since some farmers prefer free trade in cereals believing that
would make it easier for them to purchase cereals. More generally, policies have to be
designed to increase the productivity of rural residents both in on- and off-farm activi-
ties.

Consumption preference data suggest households prefer a varied mix of cereals
over a single cereal in their diet; so long as markets for cereals remain uncertain, this
suggests it will be difficult to induce farmers to gow only a single food crop, such as
maize. It was also argued that the introduction of coarse gain processing technology
and market information services could have high pay-offs in stimulating sales of coarse
gains and in reducing purchases of imported rice. These issues are discussed further in

Ch. VI.

CHAPTER VI
IMPLICATIONS FOR PUBLIC POLICY AND FUTURE RESEARCH

The Government of Senegal has set the objective of achieving 80% food self-
sufficiency by the year 2000. Policy instruments initially chosen to achieve this objective
include: (a) a producer floor price designed to stimulate the production and sales of
local coarse gains; (b) sales of coarse gain fertilizer and as of yet undeveloped fertili-
zer-responsive, improved coarse gain seed varieties; and (c) reduced sales of peanut
seed on a credit basis. Underlying each of these instruments of public policy were
certain key assumptions about the behavior of rural food system participants. These
were tested using empirical data collected during 1986/87 on 40 private traders in 7
market places and up to 210 farm households in 15 villages located throughout the
agiculturally important southeastern part of Senegal. Two complementary policy instru-
ments were the liberalization of coarse gain markets and the privatization of the
distribution of the improved inputs for cereals production mentioned in (b). This final
chapter summarizes results of the analyses presented earlier and provides selected policy
recommendations designed to assist in answering the following questions.

1. How can marketed surpluses of coarse gains in southeastern Senegal be
stimulated, at reasonable prices to consumers, while also raising rural incomes and
assuring rural food security? Are fertilizer and improved seed distribution the most
cost-effective policy options?

2. How important is the cash crop-food crop "trade-off“ (GOS policy statements
and work by other researchers indicate it is important), and how does the "trade-off“
manifest itself? A related question is, how easy will it be to induce farmers to substitute
maize for millet/sorghum production? What should national crop research priorities be

(or not be)?

209

210

3. What is the significance of, and what are the consequences of, coarse gain
market uncertainty? To what extent is a floor-price policy cost-effective and equitable as
an instrument for raising rural incomes and sales of coarse gains?

4. What are the prospects for, and anticipated consequences of, market liberali-
zation in Senegal and, more generally, Africa? Will the private sector be able to fill the
vacuum left by state withdrawal from the agicultural marketing system?

The following sections may suggest that Senegal’s rural development problems
can be tackled on a piecemeal basis, i.e., by influencing input and output markets
separately (or independently). It is important to stress that an holistic approach is
required, paying attention not only to input markets in general and specific inputs in
particular, but also to the output markets that affect the profitability of employing the
inputs.

6.1. SUMMARY AND IMPLICATIONS

6.1.]. Agicultural Input Distribution Policy

[1] Credit as a Constraint. Making credit available to households for food
purchases before and during the gowing season is likely to have high pay-offs both in
terms of raising the productivity of labor (increased energy expenditure, less time spent
searching for food), and increasing the size of the labor force. The availability of food at
the household level during the gowing season is a prerequisite for retaining (married)
sons and attracting non-family labor. Multivariate analyses show households which have
additional labor also seeded a larger quantity of coarse gains, the output of which in
principle exceeds the food needs of the additional workers and their dependents.

That food was a constraint at the beginning of the 1987 rainy season-following a
year of normal rainfalluis evident from the fact that one-half of the households sampled
would have purchased food if they had had additional funds. Private sector credit,

obtained by one-third of the households, was used primarily to purchase food. Particu-

211
larly in the southern areas, some households (15%) appear to be so constrained for food
that they have difficulty carrying over sufficient coarse gain seed from one production
season to the next.

[ii] Agricultural Equipment Supply and Maintenance. Tabular analysis reveals
households using draft technology on average produce twice as much output per farm
worker compared with households not using that technology, depending on the crop and
region. An important challenge is to expand and to sustairn the ownership of equipment
through time-i.e., to reduce the incidence of distress salesaand to promote equipment
repair facilities in rural areas. The southern survey area (Upper Casamance) was largely
ignored by the equipment distribution component of the W W: only 28%
of these households sampled were fully equipped (owning both a plow and/or a hoe and
animal draft power), in comparison to 72% in the Peanut Basin. Due to borrowing,
nearly all households in the northern area had access to equipment.

In both areas, households headed by younger individuals, owning cattle, and not
pursuing off-farm activities owned more equipment per active worker than households
without these characteristics. This is perhaps explained by the fact that younger heads
are more willing to take on the risk of owing equipment, and that on-farm investments
in livestock and equipment go hand-in-hand. This tabular analysis does not suggest that
households have to pursue off-farm activities if they are to raise their equipment
ownership level per active worker, however. Cash crop sales and association with
parastatals appear to serve that purpose.

Equipment use and availability are one important factor explaining why the
northern survey zone out-produces the southern zone, despite its lower rainfall. Promo-
ting labor-saving technology rather than fertilizer is a priority in the southern region with
its more fertile soils. Making more equipment available to more households could also

improve the distribution of rural incomes. SODEFITEX was the sole source of new

212
draft equipment in southeastern Senegal during the survey period (households in the
south working with SODEFITEX in 1986 owned more equipment per worker), but it was
drawing on limited and deteriorating stocks left over from ONCAD.

Equipment provides further advantages relative to fertilizer in that it is durable
(i.e., it can serve as collateral in future [private sector] transactions and it remains even
when the rains fail). In contrast to fertilizer, this input can also yield local off-farm
income multipliers in terms of spare parts manufacturing and equipment maintenance,
and it therefore concords well with a broader rural economic development strategy.

[iii] Fertilizer. The evidence collected in this study suggests that farmers per-
ceive the use of fertilizer on coarse gains to be profitable in a year of average rainfall
(with an average 75% yield increase) and under official prices, and that fertilizer is seen
as a necessary input for field maize production in the northern areas. Nevertheless,
nearly half of the northern farmers indicated they would not use fertilizer and improved
seed at respective prices of 100 F/kg and 210 F/kg, even if they were provided on credit.
Crude estimates suggest that most households (90%) would adopt fertilizer only if its
price fell to 70 F/kg (assuming coarse gain prices of 70 F/kg).

Farmers did not believe that maize is more fertilizer-responsive than millet and
sorghum, as is generally reported in fertilizer trials for West Africa. Furthermore, about
one-third of the farmers were convinced that animal manure is superior to mineral ferti-
lizer. This is due to their belief that manure remains in the soil for multiple years (46%
of the responses), but also because it leads to a higher yield response (40%). Not
surprisingly, farmers located in the higher-rainfall, southern research areas were more
likely to believe mineral fertilizer was better than manure than their counterparts who
were located in the drier, northern areas.

The profitability of using fertilizer on cereals will be severely squeezed if the cost

of its delivery initially rises in a privatized market and coarse gain prices fall as sales

213
rise even marginally. This requires a judicious use of output-market policies, given the
uncertainty of coarse gain markets in Senegal, as discussed further below. F urther-
more, the expected profltability of using mineral fertilizer on maize has to be analyzed
jointly with investment alternatives of farmers. At current official input / output prices,
more than 50% of the northern, but only 10% of the southern, farmers reported they
had superior investment opportunities to using fertilizer on maize. More generally, there
is a need for raising the on-farm productivity of coarse gain production. A related issue
for agonomic research concerns the potentially beneficial interaction between animal
manure and mineral fertilizer in sandy soils.

Weed control chemicals are inputs complementary to fertilizer to be considered
in Senegal’s input distribution policy. All farmers in the southern area using chemical
herbicides (obtained from SODEFITEX) were convinced its use was “worthwhile". More
research is needed on the quantitative importance and prevalence of harvest failures
caused by uncontrolled weed gowth. Limited informal observations suggest that weed
gowth may also affect farmers’ willingness to adopt fertilizer. A related finding from a
linear regession is that households having retained married sons are more likely to use
fertilizer than households without that characteristic (controlling for participation in
SODEFITEX’s cotton and maize production progam). This tends to confirm the
Delgado-McIntire (1982) hypothesis, that because of labor-shifting effects, farmers will
not adopt some improved technologies unless they also have access to additional labor.

[iv] Summary. Input market policies are integal to a strategy of stimulating
productivity gowth in southeastern Senegal. Yet different households require different
inputs, and the productivity of individual inputs on different crops varies according to
rainfall conditions and not only across but also within different regions of the country. A

pragnatic agicultural production support project would promote a broad set of inputs,

214
such as draft equipment and weed control chemicals along with fertilizer, and allow
farmers to select the inputs they expect will be most profitable for them.

The provision of spare parts and artisanal equipment maintenance should be
considered in conjunction with the objective of raising the productivity of off-farm
activities (see below). Potential new firms with econonnies of scope in providing
alternative low-cost services to farmers should be identified and promoted. Examples
may include firms selling draft equipment, maintaining the equipment, and training draft
animals. Over the longer run, local small scale manufacturing of equipment for local
conditions (such as heavier soils in the Casamance requiring more robust seeders than
the sandy soils of the Peanut Basin) could be stimulated.

6.1.2. National Crop Research Priorities

While analysts and decision makers generally view the cash crop-food crop
relationship as a competitive one, this study illustrates that when the market for one of
the crop types is thin and uncertain, there can be non-technical complementarities
among inputs. Further, cost-reducing economies of scope were found among households
producing both food and cash crops. It is consequently argued that the food crop-cash
crop trade-off in southeastern Senegal is not as severe as commonly believed, and that a
policy of only funding research on food crops would be misguided.

[i] Labor-Peanut Seed Linkage. When the household is analyzed as a coalition
of individuals with divergent interests, arising in response to high transaction costs inn
uncertain markets--instead of a black-box with a singular objective of maximizing profits-
-an interdependency emerges between cash and food crop production. Household heads
with access to peanut seed through credit from SONACOS are able to attract seasonal
workers and/or to retain married offspring within their households, in turn permitting
them to gow more food cr0ps (i.e., to allocate more seed, land and labor to food

production). So long as market uncertainty renders coarse gains unsuitable as cash

215
cash crops, reduced peanut seed distribution on credit may lead to lower rather than
higher national food production.1

[ii] Economies of Scope. A technological food crop-cash crop complementarity is
evident from the 25 % cost saving achievable when cash and food crops are produced
within the same household as opposed to crop specialization in two separate house-
holds. This result depends on the presence of a non-congested or "public" input2 (such
as labor), which arises from the fact that certain cultural operations for cash and food
crops are carried out at different times irn the agicultural calendar.

The econometric analysis of household cost functions also revealed that house-
holds are producing beyond the point of minimum average cost for food crops. At the
same time they are producing a lower amount of cash crops than that corresponding to
the level at which average costs of producing cash crops are minimized. It is thus argued
that the uncertainty of coarse gain markets is an incentive for households to produce
food beyond the minimum cost level, given current technology and prices. Given these
findings, forcing households to increase food crop production at the expense of cash crop
production would on average lead to production at points of operation even further away
from minimum average costs in the short run. The implication is, of course, of higher
food costs for buying households.

[iii] Sectoral Complementarities (Infrastructure). Aside from micro-level, on-
farm complementarities among crops, there are also complementarities between food

crops and cash craps at the sectoral level. The parastatal SODEFITEX, for example,

organizes farmers into producer associations, provides literacy progams in rural areas,

 

1A further complementarity, important particularly in the drier areas of the Peanut Basin,
is that between peanut hay required to sustain a horse during the dry season, and the
increased size of the millet field that can be cultivated by the horse.

zlhe input is non—congested or "public” in the sense that it is not fully employed at all
times (i.e., congested) if only one crop is gown.

216
erects storage facilities for coarse gains, and constructs roads providing access to rural
areas (in some instances it would have been impossible to access the survey villages
throughout the year without these roads). It is not obvious how these indirect benefits
to production in the food sector can be quantified, but the evidence for sectoral comple-
mentarities in Senegal generally corroborates similar findings from Mali (Dioné, 1989).

[iv] Food-Cash Crop Marketing. If there are complementarities between cash
crops and food crops at the input marketing and production levels, the relationship is
somewhat competitive at the output marketing level. Households selling more cash
crops are more likely to sell less food crops, m m. This suggests that once
cash needs are met through peanut and cotton sales, households may have less pressure
to sell off food stocks. As in the case of production incentives, the challenge for policy
makers in the short run is to stimulate household (production and) sales of both food
crops and cash crops, so that the economies of scope are realized. As discussed below,
policies other than those directly affecting relative food prices are available to stimulate
marketed surpluses of coarse gains.

[v] Summary. Under existing technology and prices, the food crop-cash crop
trade-off is less of a problem than commonly imagined when it is viewed from a cost
mirninization perspective. There is a cost saving, or certain resources fixed to the farm
in the short run are more fully employed, when a household produces both cash and
food crops. This is due to the existence of non-congested (i.e., slack) inputs such as
labor or equipment. Furthermore, coarse gain market uncertainty was argued to lead
to an interdependency between peanut seed, the amount of labor available to household
heads and national food output. Improving the system that provides households with
access to peanut seed is therefore important; the seed does not necessarily have to be
delivered through SONACOS-private sector options (traders, on-farm storage) should

also be explored.

217

In the short-term, reduced peanut seed availability could lead to a smaller labor
force and lower rather than higher national cereals production. Increasing the relative
profitability of cereals production, and the reliability of cereals markets, are two con-
ditions (among others) necessary for reducing this interdependency between food and
cash crops. More generally, future research should be directed toward improving the on-
farm productivity and competitiveness of food crops and cash crops, and not focus
exclusively on one type of crop or the other. This includes aggessively seeking new
market outlets for cash crops and their by-products (see also Lele, 1989, who argues for
the creation of a CGIAR-type cash crop research institute).

6.1.3. Intro-Cereals Substitution

Senegal’s New Agicultural Policy also envisions that farmers will substitute
(allegedly) more fertilizer-responsive maize for millet-sorghum production. This section
examines production, transformation and consumption issues identified from the survey
data, and related analyses of substitution and the potential expansion of maize produc-
tion.

[1] Maize Production Constraints. Maize is traditionally cultivated near the
compound (in 95% of the households sampled), where it benefits from night soil, animal
manure and close supervision. It serves primarily as an early-maturing food security
crop. The potential for expanding this type of maize production is limited, and output
increases are likely to be possible only by shifting production to field crops.

In the northern survey area the availability of mineral fertilizer poses a constraint
to extensified maize production, along with the relatively high moisture requirements of
existing varieties. In the southern area (with more fertile soils), wildlife attacks pose the
most important constraint. Maize is particularly vulnerable as it is among the first cr0ps
to mature in the gowing season. In both areas the labor-intensive nature of maize

cultivation (e.g., plowing) constrains the output of this crop. Overall, an integated

218
package, including more drought-resistant varieties, insecticides, fertilizer and labor-
saving technology should be provided if maize production is to be expanded. Also,
resource management issues arise in the south with regard to wildlife and agiculture.

[ii] Processing and Transformation Issues. Given the subsistence-orientation of
farm households, an assessment of the prospects for substituting maize for millet and
sorghum production must include transformation and cooking requirements of these
cereals (as well as consumption preferences, as discussed in [iii] below). Higher
processing and cooking requirements of maize place it at a disadvantage relative to
millet, sorghum and rice.3 Imported rice is preferred, for example, because it requires
no pounding labor relative to local coarse gains. As discussed further below, house-
holds with access to coarse gain transformation technology were less likely to purchase
rice; this may reflect reduced preparation costs of local cereals. Policy makers need to
consider the post-production tasks involved in transforming maize into an edible form in
considering prospects for intra-cereal substitutions.

[iii] Consumption Preferences and Tastes. Farmers in southeastern Senegal
prefer a dietary variety of coarse gains, which includes mixing of maize into traditional
millet and sorghum dishes. Respondents on average indicated that their diets would
comprise 33% rice, 27% millet, 22% maize and 18% sorghum if income were not a
constraint and they had not produced any coarse gains themselves. Millet and sorghum
are preferred especially during the gowing season because farmers believe they ”remain
in the stomach for a longer period of time." Rice, on the other hand, is preferred for
special occasions (ceremonies) and guests. So long as coarse gain markets are less-
than-reliable, it will therefore be difficult to induce farmers to specialize in the produc-

tion of a particular coarse gain (such as maize). Only when reliable markets provide a

 

3The opposite argument is made in southern Africa regarding millet and sorghum.

219
variety of different cereals at reasonable prices will farmers be able to specialize in the
production of a single cereal, and still maintain a dietary variety.

The data also suggest that households will switch their consumption back from
rice to local coarse gains, for example, when a year of good own-cereals production
follows one of poor production. This finding is in contrast to results reported for urban
areas in West Africa, which reveal a low price elasticity of demand for rice by poor
urban households (see, e.g., Reardon et al., 1989).

[iv] Summary. In assessing the response of rural households to a policy designed
to change the mix of cereals production, relative production and transformation or
processing costs both across and within different coarse gain types must be taken into
account, along with consumption preferences and tastes of different ethnic goups. It is
not as straightforward as assumed under the NAP to induce farmers to substitute maize
for millet and sorghum production. A further issue, not addressed here but found to be
of importance in countries such as Malawi, is the storability of different cereals and
improved cereals varieties.

6.1.4. Output Market Reform

Using evidence of observed market prices in a poor and good rainfall year, and
farmer responses to survey questions, coarse gain markets were argued to be thin and
uncertain. This feature of coarse gain markets is important since it affects the behavior
of farm households in three major areas: (1) in input markets, household heads are
induced to form coalitions with their off-spring and to use peanut seed to attract non-
family labor. Because coarse gains cannot serve as a reliable cash crop, peanut seed is
necessary to attract additional workers, who in turn allow the production of more food;
(2) in terms of crop mixes and production levels, coarse gain market uncertainty
provides households with an incentive to gow both food and cash crops to meet

household food and cash needs; and (3) in output markets, high transaction costs reduce

220
the propensity of households to participate in coarse gain markets. Households with
off-farm income activities (interpreted as providing better market information and
therefore lower transaction costs), are more likely to participate in food markets, as
buyers or sellers, ootorjo MA

[l] Limits to a Floor Price Policy. The potential for using a floor price policy to
stimulate the production and sales of coarse gains in the short run is limited. Even in
an (on average) food surplus area in a year of normal rainfall (such as the north), 20%
of the households were net buyers of food, while 21% did not participate in coarse gain
markets. The official floor price raised the cost of food to buying households, reducing
their real incomes, and it is not evident that they were able to respond positively to that
policy, given poor (private) credit markets. Moreover, net sales of coarse gains were
concentrated in a relatively small proportion of households (80% of total sales by volume
being effected by 15% of all households), implying only narrow distributional benefits to
the floor price policy.

In the southern area, where less than 10% of the households were estimated to
have been net sellers of coarse gains and about 65% of the households failed to
produce sufficient amounts of cereals (including rice) to last for one year, the floor price
was not relevant since only small quantities of local coarse gains found their way into
market places. Observed market prices (where they existed) were close to or even above
the official floor price. There was thus no leverage point for government price policy in

this region.5

 

4Another effect of off-farm activities may be that the steady income stream which they
provide increases the willingness and ability of households to manipulate their food stocks.

5 A further problem with the floor price policy is that coarse grains produced in the Gambia
were attracted into Senegal. Hence a minimum of cross-country price coordination is
necessary if unexpected effects of price policies are to be avoided.

221

For a variety of reasons, most household heads (75%) favor a floor price policy
and free cereals trade, but they disagee on a desireable price level (as expected, those
not seeding an area to cereals in 1987 sufficient to cover household consumption needs
favored a lower price, which in addition was lower in the northern than in the southern
areas). Some farmers favor a floor price policy because they believe cereals would be
more readily available. Nevertheless, over two-thirds of the household heads, including
many in the northern areas who had experienced floor prices in 1987, indicated they
would not change the area seeded to cereals if the floor price were abolished, raised or
lowered.

[ii] Alternatives to Direct Price Policies. Price policies are preferred by decision
makers since they appear to be tangible and straightforward as instruments for raising
marketed quantities of coarse gains. However, indirect price policies, which may be
more cost effective and equitable in the long run in stimulating production for sale by
changing the opportunity cost of holding food stocks and participation in markets, should
also be considered. Multiple regession analyses suggest the availability of mechanical
coarse gain transformation technology reduces purchases of rice and also stimulates
sales of coarse gains. Conceivably, the technology has this effect because it changes the
relative opportunity cost of transforming coarse gains into an edible and/or marketable
form. Providing more transformation technology to more farmers, is likely to have high
pay-offs, although more research is needed to sort out supply-side (threshing) and
demand side effects (milling). As suggested in the introduction to this section, it appears
that households with more market information are more likely to participate in coarse
gain markets because of lower transactions costs. The benefits of market information
(as a public good) need to be weighed against the cost of providing it in geogaphically

isolated areas.

222

[iii] Beyond Crop Policies. The need to pay head taxes to the State was cited as
the most important first reason (36% of the responses) for selling cash crops. The value
of taxes paid exceeded 20% of the goss value of all crops sold in 10% of the households
sampled. To the extent that taxes reduce households’ ability to invest in equipment,
Senegal’s rural tax policies may need to be re-evaluated (see Dione, 1989 for a more
thorough discussion of this issue in Mali). Since farmers sell peanuts at government-
subsidized prices to pay rural taxes, the question arises whether both taxes and subsidies
should be reduced, and what the consequences would be for different types of farmers.

More generally, the data reveal that households pursue a variety of strategies
beyond crop and/or food production to obtain income and/or achieve food security,
including the pursuit of off-farm activities (in 65% of the households) and the raising of
livestock. This in turn requiresuand provides the opportunity or leverage points for-—a
broader approach toward rural economic development progams, rather than a singular
crop-oriented focus, to raise rural incomes. The challenge is to attack problems on
simultaneous fronts, raising the productivity of economic activities both on and off the
farm.

6.1.5. The Privatization Debate

Most household heads (87%) indicated it is not a good idea for the government
(parastatals) to withdraw from input marketing, and many (67%) doubt private cereals
traders will be able to quickly fill the void left by state withdrawal. One-quarter of the
respondents believe farmers do not have a moral obligation to repay fertilizer credits in
the case of a crop failure. Most farmers (85%) would prefer to buy fertilizer and seeds
through a farmer organization, but sell their cereals surpluses individually (58%). This
could lead to credit repayment problems for traders. Existing _socti_or_n_§ vm agooises, in the
opinion of some farmers (56% in the north, 22% in the south), should not distribute the

improved cereals inputs unless they are reorganized so as to reduce patronism. Two

223
frequently cited organizations which could, in the opinion of household heads, distribute
inputs are L’As_sooiooioo gig; Moos Agjculteurs 1e 13 Qsamaoce and L’Association
\filLaggoigo, i.e., farmer-organized goups (see also Diagana, 1988).

As recently as the 1987 agicultural season, trader involvement in fertilizer
distribution was minimal in southeastern Senegal. Not surprisingly, cereals traders are
unaware of fertilizer doses, compositions and timing of application for the soils of the
regions they work in. SODEFITEX extension agents living in villages provide such
information to farmers in their progams. This raises the questions, who will provide
such knowledge to farmers under privatization?

Traders did sell equipment, draft animals and fungicides to farmers on a cash
basis during the 1987 dry season. Why are these products purchased with cash, while
fertilizer is not? On the one hand, fertilizer has not been available historically for
distribution by private traders. On the other, equipment and draft animals are more
durable and can be sold for food if there is a drought. Once applied, fertilizer is gone
(except for marginal second-year benefits) and cannot be resold to buy food. Fungicide
has a more or less proven record of effectiveness for farmers, is relatively inexpensive
(and more necessary in the farmer’s mind?), and is used on a crop which has assured
market outlets at known pricey-peanuts. This partially explains why there is effective
cash demand for these inputs in the areas, but not for mineral fertilizer.

Further, market prospects for maize in rural areas are weak. This is due to
consumer preferences for other cereals, problems of home-processing and storage of
maize, and the fact that most farmers gow their own maize for consumption during the
gowing season. A potentially important role for the public sector is to fund research
and pilot projects to expand maize utilization (for example, consider joint ventures
between commercial food manufacturers and private traders). Demand for maize could

be stimulated by developing new processing facilities, introducing more efficient stoves,

224
new product forms (especially for urban areas), and possibly using maize as livestock
feed. Again, these developments are unlikely to materialize without some initial
guidance and support from public sector research and development organizations.

A resilient private sector in Senegal appears potentially capable of carrying out
more functions as envisioned under the NAP. Private actors are already selling certain
types of agicultural inputs, and seeking out profitable trading opportunities (peanuts
and their derivatives, consumer goods, Gambian rice and fertilizer). Morris (1988) has
demonstrated for the Fleuve that the private sector can effectively compete with a
parastatal if certain market-wide preconditions exist, and if private agents are not
expected to bear all of the initial risks of a given activity (such as developing processing
equipment). The same argument is likely to apply to southeastern Senegal.

An important privatization issue is not so much the question of whether the pri-
vate or the public sector is ”better” (i.e., more cost-effective) at supplying a certain set of
services, but how the two sectors can best complement one another in providing the
entire set of services required by producers and traders. While the record of dismal past
parastatal performance cannot be ignored, not all parastatals are alike and important
lessons about what does and does not work in terms of input distribution and output
marketing can be learned, for example, from the SODEFITEX-model (including how
they identify farmers with whom to work, how they organize the farmers into goups,
and how they deal with debt repayment problems in the case of crop failures).6 At the
same time simple analogies across crop sectors must be drawn with caution; SODEFI-
TEX operates with the considerable advantage that cotton cannot be consumed on the
farm, which facilitates credit recuperation. Raising the on-farm productivity of cereals is

one way of reducing default on cereals loans and increasing private sector involvement.

 

68cc Diagana, 1988, for a detailed discussion of how SODEFITEX organizes producers.
Also Ba, 1986.

225
6.1.6. Privatization and the Interaction Among Technology, Prices and
Institutions.

Some southern household heads indicated they would have saved rather than
invested additional funds in agiculture in 1987. This response reflects limited perceived
and actual investment opportunities due to a lack of labor and /or equipment markets
combined with unreliable (food crops) or limited (cash crops) output markets (the latter
due to limited inputs available from parastatals and/or for purchase from other produ-
cers in the case of peanut seed).

This suggests that there is a general coordination failure in the input-output
markets surrounding farmers in southeastern Senegal. A counter-example case study
shows that successful coordination can lead to high performance levels. One example is
a farmer (also a village chief and religious leader) in the northern area who operates as
follows: He works under a contract with the cotton parastatal SODEFITEX, which
assures input delivery, an output market (for cotton and maize), and the presence of a

SODEFITEX extension agent in the village. He uses the following technology package.

0 a cotton-maize rotation
o plowing to the correct depth at the right time
a mineral fertilizer application at the right time, (using a dose somewhat

higher than that recommended by ISRA)
o mechanical seeding and correct (intra- and inter-) row spacing
o herbicide applied at the right time combined with manual and mechanical
weeding later in the season,
This individual was able to produce 4000 kg/ha of maize in one season, representing 3-4

times the average regional yield.7

 

7The individual also had access to a travelling equipment repair man. The village was
located at a distance of approximately 15 km from Ndoga Babacar, which constituted a

226

The argument is not that all economic activity has to be organized or coordinated
by the public sector. However, a parastatal such as SODEFITEX has the economies of
scale and size enabling it to provide an extension agent to relatively remote villages, as
well as roads to access those villages. Without these ”public" goods, which rational and
self-motivated businessmen will not provide (even if they were able to) since they can
not exclude other individuals from benefiting from such investments, private actors either
will not become significant agents of change or they will do so at prohibitive costs.
These costs will be passed on to farmers and consumers, unless productivity can be
raised elsewhere in the food marketing Chain in compensation.

A similar problem involves the supply of and demand for local coarse gains.
Due to high transport costs and the relative isolation of rural areas, which together cause
a subsistence-orientation in food production (see also Binswanger and McIntire, 1987),
food supply and demand are poorly articulated by private, self-motivated actors: in good
rainfall years, when supplies are high, demand is low and selling of food is not remune-
rative. The opposite obtains in poor rainfall years, with food becoming unaffordable.
The average result is continued subsistence production with little incentive for invest-
ment in improved cereals technology-which is turn is interdependent with unavailability
of the technology. Mechanisms are needed to coordinate and/or assist in managing food
supply and demand over time by providing incentives to: (a) conduct storage functions
(temporal arbitrage between good and bad years); (b) complete improved spatial
arbitrage between food production deficit and surplus regions through reduced transport
and information costsuthe latter will be effective in stabilizing output markets if yields in

different regions of Senegal are not co-variant; and (c) increase but stabilize demand in

 

constraint to the marketing of cereals through private channels; however, the village was
visited by a truck sent by SODEFITEX during the marketing season. As a reward for
gowing 8 ha of maize in 1985, and 9 in 1986, in his personal field, this individual was
provided with an airplane ticket to Mecca by SODEFITEX.

227
food deficit areas over time by lowering costs of local cereals to consumers (through
better processing, transport, etc.).
Under current conditions there is little rural specialization into agicultural and
non-agicultural activities, which tends to be both caused by and a cause of unreliable

output markets.8

Until specialized non-farming firms (and urban areas) provide
agicultural households with reliable sources of food demand, the presence of the CSA
or some such demand-creating entity is necessary (but not sufficient) to stimulate coarse
gain production. Higher coarse gain prices alone will not lead to the production of
food surpluses among farmers barely able to meet subsistence requirements. An output
price policy has to be complemented with an input package and a policy, or the develop-
ment of a (credit) market, which relaxes resource constraints and provides some of the
expected benefits accruing at harvest as early as the preceding planting season (so that
the inputs can be purchased and used at that time). Obviously, increased marketed
surpluses can lead to severely depressed output prices in the post-harvest season if
output markets are not sufficiently integated. This in turn affects the financial profita-
bility to farmers of new technologies such as fertilizer. This is yet another instance of
the important interaction among markets/ institutions and technology.
6.2. LIMITATIONS OF THE STUDY AND DIRECTIONS FOR FUTURE RESEARCH
To some extent this study was handicapped by being carried out too early relative
to actual policy changes in southeastern Senegal, an area about which little empirical
knowledge existed previously. It was originally expected that changes in the input
distribution system would take place before the study was completed, permitting a "pre-

and post-NAP“ analysis of farmer (and trader) behavior. In fact, parastatals continued

to be the main actors in terms of input distribution (SONACOS’ withdrawal from peanut

 

85cc also the argument in Shaffer et al., 1983.

228
seed sales on credit in 1987 being an exception). Nevertheless, the analysis of answers to
hypothetical questions, aside from confirming manifest behavior, provided useful insights
for evaluating the feasibility of certain policies in effecting economic gowth in the area.
One finding is that the private sector cannot be expected to immediately fill a vacuum
left by parastatal withdrawal.

The study did not have the resources to examine anticipated intra-household
effects of the NAP (other than superficially in terms of seed and chemical input allo-
cation). Hence it is impossible to predict precisely how changes in technology and crop
mixes will affect the welfare (income) and productivity of certain household members
such as females.9 These changes can be as subtle as a woman losing her remuneration
for carrying millet from the field to the home when the household head adopts a cart
and draft animal and carries out the task himself (Gastellu, 1988, p.125), or as obvious
as in situations where men take on the gowing of a crop following the introduction of a
labor productivity-increasing technology (cf. von Braun, 1988, p.1095, for the case of rice
in the Gambia).

The data collected provided some insight into food (cereals, and peanut)
availability at the household level, but nothing could be said about the nutritional status
of individual household members such as children. Some research suggests that access
to kilocalories through cereals is positively correlated with adequate protein intake (see,
e.g., Sukhatme, 1970). Collection of consumption and transactions price and quantity
data for individual household members and food items would provide important insights

in this regard.10

 

9For example, Diop, 1985, p.167 writes that peanuts provide women with economic
independence (from their spouses).

10In one southern market (Sare Yoba Diega) market transactions were insignificant,
however.

229

Detailed crop budget information is required to understand the on-farm tech-
nological relationship between (cash and food) crops, and improved inputs such as
fertilizer, under different soil types and rainfall conditions. Data on input allocation by
crop (combined with seasonal consumption data) would also permit examining the
effects of food consumption during the gowing season on the productivity of labor (see,
e. g., Strauss, 1986); studying the relationship between technical-scientific knowledge of
farmers and the profitability of improved technolog'; and evaluating the changes (if any)
in productivity needed to make cereals more competitive with traditional ”cash crops”.

This study was also unable to investigate in sufficient detail the effects of off-
farm activities and livestock ownership on household welfare. More research is needed
on the benefits and costs of these alternative income-generating activities, the skills and
knowledge required to pursue them, and their effects on the willingness and ability of
households to invest in improved inputs. Results from such findings need to be related
to increased specialization, productivity and employment in rural areas of Senegal.

As this study has shown, it is not enough to focus only on on-farm production of
cereals, if the welfare of rural residents is to be increased. More research is needed on
the benefits and costs of related activities which add value (utility) to the product, such
as storing, processing, cooking and transportation of cereals, and the role of tastes in
driving production and consumption.11 Imported (broken) rice is favored not necessa-
rily because it is viewed as a luxury item but because it offers real economic advantages

in terms of preparation costs. 12

 

11Millet cannot be stored, for example, once it has been processed. This reduces the
incentive for employing mechanical transformation technology located at some distance from
one’s residence. One progam strategy could be promoting the development of small-
scale, low cost technology which would be made available to all villages.

12Ross (1980) discusses problems involved in encouraging urban consumers (Dakar) to
switch from imported rice to millet and maize (and local rice). He also finds that
consumers with higher incomes are willing to pay a premium for the Siam rice variety,

230

A final limitation of this study, as is true of all short-term efforts, is that it
covered only a 12-month‘ period with unique rainfall characteristics and patterns of
resource ownership and allocation. Hence the conclusions must be interpreted with
caution. Ideally, a panel of farmers would be followed over a number of years to better
understand their coping strategies and behavior.

6.3. KNOWLEDGE, PERCEPTIONS AND BELIEFS: IMPLICATIONS FOR

EXTENSION AND FURTHER RESEARCH EFFORT ON THE MARKET
REFORM PROCESS

Policy reform is not a one-time activity, although privatization is often billed as
such. Policy makers and analysts (advisors) are dealing with complex food systems and
detailed micro-level empirical knowledge is needed to test the assumptions, driven by
ideology, lack of data or both, underlying recommendations for restructuring incentives
of food system participants to increase the welfare of rural residents. In Senegal, it is
attractive but not sufficient to decree reduced peanut seed distribution on credit; to
demand increased use of fertilizer on cereals; and to mandate a producer floor price for
cereals to raise national food production and rural welfare.

Policy reform should also not be limited to a single, topical ”issue" in the input-
production-marketing-consumption sequence. Increased production of food through new
inputs will not fully benefit urban consumers if transportation and processing bottlenecks
prevent food prices from falling as much as they could have in response to increased
(unprocessed) supplies. It behooves policy makers to not place all their eggs in a single
basket.

The knowledge generated in this study contributes to an understanding of the

possible effects of new policies, and how food system participants may or may not

 

suggesting future analyses must take into consideration different types of rice.

231
respond to them. However, as the dynamic food system evolves under a new set of
policies, it will be necessary to continue to monitor the effects of, and constraints to the
implementation of, economic development progams through data collection and analysis

of economic activities in rural areas.

232

APPENDIX A-l:

EQUIPMENT AND DRAFT ANIMAL PRICES

TABLE A-la: EQUIPMENT PRICES

 

 

Total Annua
Item Cost Cost

-------- in FCFA--------
Plow 20,000 6,200
Hoe 24,000 7,740
Seeder b 20,000 6,200
Soulevoose 18,000 6,000
Cart 19,000-24,000 2,100

 

Sourcg: SODEFITEX (B. Diop), personal communi-
cation, 1987 (except for sooloyoooo).

a. Purchased on credit.
b. Estimate.
c. Depending on draft animal.

TABLE A-lb: DRAFT ANIMAL PRICES

 

 

Total Useful Annual
Item Cost Life Cost

(FCFA) (Years) (FCFA)
Cattle/Oxen 57,375 5 11,475
Horse 85,000 8 10,625
Donkey 17,500 7 2,500

 

Sourco: ISRA/MSU FSP Surveys, 1986/7.

233

APPENDIX A-2:
PARTIAL BUDGET FOR HERBICIDE USE
CASE STUDY IN THE SOUTHERN AREA

The following calculations, using data from one progessive southern farmer,
illustrate the monetary trade-offs between manual weeding and the use of herbicides.

Assomotioos: Only one weeding operation or one application of herbicide are
carried out since animals were gazed in the field prior to planting and, according to the
farmer, theurine of animals killed off some of the weeds; herbicide is applied after
seeding, the manual operation is carried out 15 days after seeding.

Cash cost of 1 liter (1) of herbicide is 2,500 FCFA.
Application rate is 4 l/ha.

Credit cost of herbicide pump = 18,000 FCFA, amortized
over 5 years and 2 hectares.

Cash cost of batteries = 415 FCFA/pack of 5 batteries.
Cost of application labor = 0.5 hours valued at 114
FCFA/hour.

Cost of manual weeding = 124 hours per ha valued at 114
FCFA/hour.

TABLE A-2: PARTIAL BUDGET CALCULATIONS

 

 

Total
Expense Item CFA CFA
Cost of Manual Weeding 14,592
Cash cost of Herbicide 10,000
Batteries 415
Labor for Application 57
Cost of Herbicide Pump 1,800
Total 42,272
Net Grain (Manual-herbicide) 2,320

Gross Benefit-Cost Ratio 1.19

Soogco: ISRA/MSU FSP Surveys, 1986/7.

Note: if the herbicide had been bought on credit @ 2850
FCFA/l, the net gain would have fallen to 920
FCFA.

234

APPENDIX A-3:
FERTILIZER BENEFIT-COST RATIOS BASED ON FARMERS’
YIELD RESPONSE PERCEPTIONS

ASSUMPTIONS USED IN CALCULATING
BENEFIT-COST RATIOS

[TABLES A-3a, A-3b, AND A-3c]

1. Using 1986 yield averages by Arrondissement, which reflect the use of
fertilizer on some of the cereals (therefore there is a tendency to overestimate yields
under the "no fertilizer” situation).

2. Assumes farmer applies all the fertilizer at recommended application rates
(exception in Table A-3b).

3. Assumes official prices (except in Table 3-20 in the text).

4. Labor cost of fertilizer application is not taken into account.

235

YIELD RESPONSE PERCEPTIONS

A. 1987 RAINFALL CONDITIONS

TABLE A-3a: PARTIAL FERTILIZER BUDGET ANALYSIS USING FARMERS’

 

 

 

 

CFA/ko Output: 81ne Baloua Sen. Oriental Casamance
70 (Kounoheul) (Hake) (see note)

1. NILLBT YIELD 782 781 778
Baae Yield 782 781 778
Value of Outaut 54.740 54.670 54.460
Koa NPK + Urea 200 200 200
NPK CPA/k9 88 91 94
Urea CFA/ko 77 80 83
Total Coat 17.050 17.650 18.250
Yield a] fart. 1.228 1.218 1.486
Value of Outaut 85.942 85.285 104.019
Net Benefit 14.152 12.965 31.309
Groaa BIC Ratio 1.83 1.73 2.72
2. BORGHUH YLO. 957 905- 1.082
Baaa Yield 905 1.082
Value of Output 63.350 75.740
Koa NPK e Urea 200 200
NPK CPA/k0 91 94
Urea CPA/k9 80 83
Total Coat 17.650 18,250
Yield 0’ fort. 1.466 2.034
Value of Outout 102.627 142.391
Net Beneflt' 21.627 48.401
Groaa B/C Ratio 2.23 3,65
3. FLO HAIZE YLO 1.141 1.073 1.206
Baae Yield 1.141 1.073 1.206
Value of Cutout 79.870 75.110 84.420
Koa NPK + Urea 250 250 250
NPK CPA/k9 88 91 94
Urea CPA/k9 77 80 83
Total Coat 20.900 21.650 22.400
Yield u/ fart. 1.848 1.663 2.340
Value of Outout 129.389 116.421 163.775
Nat Benefit 28.619 19.661 56.955
Groaa B/C Ratio 2.37 1.91 3.54

 

Arrondlaaementa are: Kounkane.

ano and Oioulaoolon

236

YIELD RESPONSE PERCEPTIONS

B. POOR RAINFALL CONDITIONS
[HALF DOSE OF FERTILIZER]

TABLE A-3b: PARTIAL FERTILIZER BUDGET ANALYSIS USING FARMERS’

 

 

 

 

 

CPA/Ra Outout: 51ne Baloum Ben. Oriental Casamance
70 (Kounoneul) (Haka) (see note)

1. HILLET YIELD 782 781 778
Base Yleld 430 266 381
Value of Outout 30.107 18.588 26.685
K08 NPK + Urea 100 100 106
NPK CPA/k9 88 91 94
Urea CFA/ko 77 80 83
Total Cost 8.525 8.825 9.125
Yield u/ fart. 954 461 716
Value of Output 66.783 32.255' 50.103
Net Benefit 28.151 4.842 14.293
Gross BIC Ratio 4.30 1.55 2.57
2. BORGHUH YLD. 957 905 1.082
Base Yield 262 584
Value of Output 18.372 40.900
Koa NPK + Urea 100 100
NPK CPA/k9 91 94
Urea CPA/kc 80 83
Total Coat 8.825 9.125
Yield w/ fart. 4:17 1.017
Value of Output 28.508 71.196
Nat Benefit 1.311 21.171
Groaa B/C Ratio 1.15 3.32
3. FLO HAIZE YLD 1.141 1.073 1.206
Baae Yield 616 215 507
Value of Output 43.130 15.022 35,456
Koa NPK + Urea 125 125 125
NPK CFA/ko 88 91 94
Urea CPA/k9 77 80 83
Total Coat 10.450 10.825 11.200
Yield a] fart. 1.460 633 1.013
Value of Cutout 102.234 44.315 70.913}
Nat Banafit 48.654 18.468 24.256
Groaa B/C Ratio 5.66 2.71 3.17

Arrondlaaementa are: Kounkane. Dabo and Dloulacolon

237

TABLE A-3 : PARTIAL FERTILIZER BUDGET ANALYSIS USING FARMERS’
YIELD RESPONSE PERCEPTIONS

C. POOR RAINFALL CONDITIONS

 

 

 

 

[FULL DOSE OF FERTILIZER]
CFA/kd Output: Sine Saloua Sen. Oriental Casamance
7O (Kounoneul) (Maka) (see note)
1. HILLET YIELD 782 781 778
Base Yield 430 266 381
Value of Output 24.086 14.870 21.348
Koa NPK + Urea 200 200 200
NPK CPA/k0 88 91 94
Urea CPA/kg 77 80 83
Total Coat 17.050 17.650 18.250
Yield u/ fert. 954 461 716
Value of Output 53.426 25.804' 40.083
Net Benefit 12.291 (6.716) 484
Gross 8/0 Ratio 1.72 0.62 1.03
2. SORGHUH YLO. 957 905 1.082
Base Yield 262 584
Value of Output 14.697 32.720
Koa NPK + Urea 200 200
NPK CFA/ko 91 94
Urea CPA/k9 80 83
Yield HI fort. 407 1.017
Value of Output 22.806 56.956
Net Benefit (9.541) 5.987
Groaa B/c Ratio 0.46 1.33
3. FLO NAIZE YLO 1.141 1.073 1.206
Baae Yield 616 215 507
Value of Output 34.504 12.018 28.365
Koa NPK + Urea 250 250 250
NPK CPA/k6 88 91 94
Urea CPA/k9 77 80 83
Total Coat 20.900 21.650 22.400
Yield u/ fert. 1.460 633 1.013
Value of Outaut 81.787 35.452 56.730
Net Benefit 26.383 1.784 5.965
Groaa BIC Ratio 2.26 1.08 1.27

 

Arrondissementa are:

Kounkane.

Oabo and Oioulaoolon

238

AKPPEmflJDKAb4:
PRHDKJHUULIHIHJ)BHAJZIJPFKJDRNZFHDAIPTKJBLIEHS

TABLE A-4: PRINCIPAL GENERAL FIELD MAIZE
PRODUCTION PROBLEMS. 1987

 

 

First Response

North

 

South

 

Total

 

----% of Households----

 

 

Wildlife Predators ......... 2% 75% 39%
High Fertility Needs ....... 56% 2% 29%
Labor Intensive ............ 20% 8% 14%
High Rainfall Needs ........ 20% 0% 10%
Insect Attacks ............. 0% 14% 7%
Other Response ............. 2% 2% 2%
Total ...................... 100% 100% 100%
Second Response

High Fertility Needs ....... 38% 25% 31%
Wildlife Predators ......... 21% 29% 26%
Insect Attacks ............. 0% 24% 13%
Labor Intensive ............ 21% 18% 19%
High Rainfall Needs ........ 14% 0% 6%
Other Response ............. 5% 4% 4%
Total ...................... 100% 100% 100%
Third Response

Insect Attacks ............. 23% 38% 33%
Wildlife Predators ......... 27% 27% 27%
Labor Intensive ............ 18% 21% 20%
High Fertility Needs ....... 0% 13% 9%
Other Response ............. 23% 2% 9%
High Rainfall Needs ........ 9% 0% 3%
Total ...................... 100% 100% 100%

 

Source: ISRA/MSU Food Security Project Survey, 1987

239

APPENDIX A-S:
PRICES UNDERLYING THE COST DATA

TABLE A-5: PRICES 0F VARIABLE INPUTSa

 

 

FCFA per
Item Unit Unit
N-P-K Fertilizer kg 88
Urea Fertilizer kg 88
Insecticdes liters 40
Herbicide kg 2,700
Fungicide kg 300
Coarse Grain Seed kg 70
Rice Seed kg 130
Peanut Seed kg 90
CottoB kg 100
Labor AEP-hrs 98.5

 

Source: ISRA/MSU FSP Survey, 1987

a. See Appendix A-l for fixed costs (equipment
prices) and draft animal prices.

b. Wage rate varies by region; labor is assumed
to be employed for 10 months, at 30 days
a month and 8 hours per day.

240

APPENDIX A-6:
NOTE ON THE "FORCED SALES” HYPOTHESIS

15 households both sold and bought coarse gains during the survey period.

Of these 15 households, 12 were in the north and 3 were in the south. Since the
northern areas are more ”commercially active," this suggests the northern households

were possibly already responding to the New Agicultural Policy, i.e., engaging in cereals
trade.

Period 1 Period 2
(Post Harvest) (Pre Planting)
L L J

 

Note on time periods: I y I
Oct/Nov Feb July

Of the 12 northern households in the buy&sell category,

5 households sold 74 kgs during period 1 and
bought 109 kgs during period 2.

1 household sold 20 kgs and
bought 50 kgs, both in period 1.

1 household sold 350 kgs and
bought 150 kgs, both in period 2, and

5 households appeared to be trading within or across the two periods, i.e., they
first purchased coarse gains and then sold coarse gains (all but one of these
households--buying 82 kgs in period 1 and selling & buying 400 kgs during period 2 were
net coarse gain sellers).

For the three southern households in the buy&sell category,

1 household bought 200 kgs and
sold 250 kgs, both in period 1.

1 household bought 284 kgs in period 1 and
sold 47 kgs and bought 234 kgs, both in period 2.

1 household sold 280 kgs in period 1 and
bought 14 kgs in period 2.

From the above, and on the basis of the small number of households involved, it
is concluded that the hypothesis of ”forced sales" after harvest and repurchases later on
in the season at higher prices generally does not hold for this sample.

241

APPENDIX A-7:
RESERVATION PRICES FOR SELLING CEREALS

TABLE A-7: HOUSEHOLD HEADS’ RESERVATION PRICES FOR
SELLING THEIR OWN CEREALS, JUNE 1987

 

 

North South

 

 

Da ss 5 D 55 s Totalb

 

Mean FCFA 83 94 101 113 107c 99 104
Std. Devn. 24 21 25 46 28 28 34
Valid N. 4 24 45 so 23 6 152

 

§ouzggz ISRA/MSU FSP Survey, 1987

a. D=Deficit, SS=Self~sufficienct; S=Surplus area seeded to
coarse gains in 1987 (according to farmer’s assesment).

b. For 154/208 (=74%) farmers willing to give a response.

c. Calculation excludes 2 outliers of 500 FCFA/kg.

BIBLIOGRAPHY

Abt Associates, Inc. Senegal Aggicultugal Policy Analysis Agicultural Policy Analysis
Report, Cambridge (MA), April 1985 (mimeo).

AgoProgess Report. "Etude de Factabilité pour l’Equipement des Magasins de
Stockage.” République du Sénégal, MDR, CSA, Rapport Définitif, Novembre
1970 (mimeo).

Akridge, J.T. and T.W. Hertel. "Multiproduct Cost Relationships for Retail Fertilizer
Plants." Anger. ,lml, Agzic. Econ. 68(4)1986:928-38.

Alameda, 1.1. and AJ. Mann. "Energy Price Shocks, Input Price Changes and Develop-
mental Implications: A Translog Model Applied to Puerto Rico." Jml. D_evel.
SM. 25(3)1989:327-43.

Antle, J .M. "The Structure of US. Technology, 1970-78." AW.
66(4)1984:414-21.

Axelrod, RM. The Evolution of Cooperation, New York, Basic Books, 1984.

Ba, C. Lgs Pgul du Sénégalufitude Géogganhjgue. Les Nouvelles Editions Africaines
(NEA), Dakar, Senegal, 1986.

Barbier, 5.3. ”Cash Crops, Food Crops and Sustainability: The Case of Indonesia."
Worlgl Devel. 19(6)1989:879-95.

Bardhan, K. "Price and Output Response of Marketed Surplus of Food Grains: A
Cross-Sectional Study of Some North Indian Villages.” Amer. ,1, Ag", Econ.
52(1970):51-61.

Barnum, H. and L. Squire. A Model of an Agicultural Household. Washington DC:
World Bank Occas. Pap. 27, 1979.

Baumol, WJ., J.C. Panzar and RD. Willig. antestnblg Markets and the lhegg of
Industly Structure. Harcourt Brace Jovanovich, Inc. New York, 1982.

Bell, C. and C. Sussangkam. ”Rationing and Adjustment in the Market for Tenancies:

The Behavior of Landowning Households in Thanjavur District.” Amer, J. Ag;
Econ. 70(1988):779-89.

Benoit-Cattin, M. and J. Faye. L’Egploitation Agg'cole Familiale en Afrigug §oudano-
Sabélienne, Presses Universitaires de France, Paris, France, 1982.

Bertrand, R. et J. Valenza. "Evaluation des Possibilités Agicoles, Pastorales et
Forestieres au Milieux Naturels du Sénégal Oriental"; Montpellier, France, 1970.

Bhagwati, J.N. ”Immiserizing Growth: A Geometrical Note.” Ch. 21 in Bhagwati, ed.,
lntegnatignal Trade: Selected Readings. MIT Press, Cambridge, MA: 1982.

242

243

Binswanger, H.P. ”A Cost Function Approach to the Measurement of Elasticities of
Factor Demand and Elasticities of Substitution." Amen, lml. Agric. Eeon.
56, 1974:377-86.

Binswanger, HP. and MR. Rosenzweig "Behavioral and Material Determinants of
Production Relations in Agiculture." ,lznl. Dev, Stud. 22(3),1986:503-39.

Binswanger, HF. and J. McIntire. ”Behavioral and Material Determinants of Production
Relations in Land-abundant Tropical Agiculture." W.
36(1),1987:73-99.

‘Bromley, D.W. and J.-P. Chavas. "On Risk, Transactions, and Economic Development in

the Semiarid Tropics." Econ. Dev. gull. Enng. 37(4), 1989:719-736.

Brown, R.S., D.W. Caves and LR. Christensen. "Modelling the Structure of Cost and
Production for Multiproduct Firms.” Sonln, Eeen. ,lrnl. 46, 1979:256-73.

Casley, DJ. and DA. Lury. Dat o ' ' eve o ' Cou ie . Clarendon Press,
Oxford, England, 1982.

Caves, D.W., L.R. Christensen and M.W. Tretheway. "Flexible Cost Functions for
Multiproduct Firms.” Rev. Econ. Stel. LXII(3), 1980:477-81.

Chacholiades, M. lntegnalienal Trade jlheog and Polley. McGraw-Hill Book Company,
New York, 1978.

Chambers, R.G. A lie oductio Anal sis: A ua A oach. Cambridge Univ. Press,
Cambridge, 1988.

Chambers, R.G. and H. Lee. "Constrained Output Maximization and US. Agriculture."

AQQILCiELOLL 18, 1986:347-57.

Cragg, J .6. ”Some Statistical Models for Limited Dependent Variables with Applica-
tions to the Demand for Durable Goods.” Econometrica. 39(1971):829-44.

Crawford, E.W. "Factors Affecting Household Food Security." In: Global Eood Security:
Focus on Africa, Study/Action Package, prepared for World Food Day, October
16, 1988, pp. B-9 to B-14.

Crawford, E.W., C. Jolly, V. Kelly, P. Lambrecht, M. Mbaye, and M. Gaye, 1987, "A
Field Study of Fertilizer Distribution and Use in Senegal, 1984: Final Report",

MSU International Development Papers Reprint, No. 11, East Lansing, Michigan
(originally published in 1985 in Senegal).

Crawford, E.W., J.S. Holtzman, J.M. Staatz, C. Wolf and M.T. Weber, 1988, ”MSU Expe-
rience in Research Design and Data Processing/Analysis"; paper presented at the
University of Zambia Food Security Research Network Workshop, Lusaka,
Zambia, May 24-25.

Cyert, RM. and JG. March. A Behavioral jljheog ef the Eim, Englewood Cliffs, N.J.,
Prentice-Hall, 1963.

244

D’Agostino, V.C. "Coarse Grain Production and Transactions in Mali: Farm Household
Strategies and Government Policy." Unpubl. M.Sc. Thesis, Michigan State
University, East Lansing, Michigan, 1988.

D’Agostino, V.C. and S. Sundberg, "Household Drought Coping Strategies in Mali,"
forthcoming Working Paper, Michigan State University, East Lansing, 1990
(mimeo).

Delgado, CL. and J. McIntire, ”Constraints on Oxen Cultivation in the Sahel," Amer.

W 64(1932)=133-196-

Diagana, B.N., ”Les Organisations Paraétatiques et Paysannes dans le Sud-est du
Senegal“, Dakar: Direction des Recherches sur les Systemes Agaires de l’ISRA,
Décembre 1987, 18p.

 

Diagana, B.N. "La NPA du Sénégal et la Responsibilisation des Organisations des
Producteurs (OPS): Participation et Opinions des Producteurs au Sud-Est du
Sénégal," Dakar: Direction des Recherches sur les Systemes Agaires de l’ISRA,
Juin 1988.

Diagana, B.N., A.A. Fall et SJ. Goetz, "Observations Préliminaires sur les Participants
du Systeme Ago-Alimentaire dans les Regions Orientales du Sénégal", presented
at the AADPA Conference March 29-April 1, 1987, Hotel de l’Indépendence,
Dakar.

Diagana, B.N. et SJ. Goetz, "Les SRD et le Systeme de Distribution des Intrants en 1985
et 1986: Participation et Opinions des Producteurs au Sud-est du Sénégal", Note
d’Information BAME 87-2, Décembre 1987, 14p.

Dione, J. ”Informing Food Security Policy in Mali: Interactions Between Technology,
Institutions and Market Reforms." Unpubl. PhD Dissertation, Michigan State
University, East Lansing, Michigan, 1989.

Diop, A.-B. a ' W0 «T adit' t an e e . Editions Karthala, Paris,
France, 1985.

Diop, B. "Rapport Semestriel", SODEFITEX Cellule Recherche et Developpement,
Tambacounda, Octobre 1986 (mimeo).

Dornbusch, R. Open Economy Macroeconomig. Basic Books, New York, 1980.

Eicher, C.K. ”Facing up to Africa’s Food Crisis," reprinted from Foreign Affaizs,
61(1982):154-74, in Aggienltnral Developmenl in the jlfnirg World, edited by CK.
Eicher and J.M. Staatz, 1984, pp. 453-79. Baltimore: The John Hopkins
University Press.

Eicher, CK. ”Food Security Battles in Sub-Saharan Africa.” Revised version of a plenary
address presented at the VIIth World Congess for Rural Sociology, June 26 -
July 2, 1988, Bologna, Italy.

245

Eicher, C.K. ”Sustainable Institutions for African Agicultural Development.“ Working
Paper No. 19, ISNAR (International Service for National Agicultural Research),
February, 1989.

Eicher, GK. and DC. Baker, ”Research on Agicultural Development in Sub-Saharan
Africa: A Critical Review." U t atio eveo men a e No. 1. Dept.
Agic. Econ., Michigan State Univeristy, East Lansing, Michigan, 1982.

Eicher. CR and JM. Staatz, edS- 8W9. The
John Hopkins University Press, Baltimore, 1984.

Fall, A.A. "Etude Descriptive des Marches Cerealiers au Sud-Est du Sénégal", Dakar:
Direction des Recherches sur les Systemes Agaires de l’ISRA, Dec. 1987, 21p.
(87-3).

FAO, ”Politiques et Progammes Céréalers", [Cereals Study], various Volumes; prepared
for the MDR/Government of Senegal (undated--1984).

Gastellu, J .-M. ”... Mais, Oil Sont Donc ces Unites Economiques que nos Amis

Cherchent Tant en Afrique?" Cahiers O.B.S.flf,Q.M., ser. Sciences Humeines,
Vol. XVII, Nos. 1-2, 1980:3-11.

Gastellu, J. M. "Petit Traité de MatrilineageuL’Accumulation dans Deux Sociétés

Rurales d’Afrique de l’Ouest.‘ Qahiers O. B. S. T. O. M. ., eeg, Scienees flnmgines s,
Vol. XXI, No.4 1.985 413- 32.

Gastellu, J.-M. "Le Paysan, l’Etat et les Sécheresses (Ngohé, Sénégal: 1972-1982).”
ganiegs des Sciences Humalnes, 24(1)1988:119-136.

Gaye, M. ”Le Desengagement de l’Etat et la Problematique des Intrants Agicoles au
Sénégal,” ISRA/DRSAER, 1986 (mimeo).

Gaye, M. "Les Sections Villageoises et le Credit." Dakar: Direction de Recherches sur
les Systemes Agaires de l’ISRA, Sept. 1987, Document de travail 87-1.

Goetz, SJ. "Prospects for Expanded Maize Production in Southeastern Senegal", Briefing
Note presented to USAID/Dakar, January 1988, 6p.

Goetz, S.J., ”Agricultural Product Prices in Southeastern Senegal: Final Report for 1987",
presented to USAID/Dakar, April 1988, 17p.

Goetz, SJ. "A Note on Off-Farm Activities and Food Security in Southeastern Senegal,"
Draft Briefing Note presented to USAID/Dakar, June 1988, 8p.

Goetz, SJ ., "Observations on Rural Self-Sufficiency and Prospects for Expanding Cereals

Production in Southeastern Senegal: A Preliminary Report”, presented to
USAID/ Dakar, April 1988, 23p. [revised Sept. 1988, ISRA/BAME publication
series].

246

Goetz, S.J., E.W. Crawford, B.N. Diagana, J .S. Holtzman and M.T. Weber. "An Analysis
of the Effects of Changes in Marketing Institutions and Policy on Cereals
Producers and Marketing Agents in Southeastern Senegal: Executive Summary”.
Michigan State University Staff Working Paper No. 88-149. East Lansing,
Michigan, 1988.

Goetz, SJ. et B.N. Diagana avec A.K. Diallo, ”Le Projet Securité Alimentaire
ISRA/MSU/USAID: Note Methodologique", Dakar: Direction de Recherches
sur les Systemes Agaires dc l’ISRA, Decembre 1987, 21p.

Goetz, 8.]. with A. Dieng, ”Characteristics of Agiculture and Farm Households in
South-eastern Senegal”, ISRA/BAME Publication Series, March 1987, 32p.

Goetz, S.J., A.A. Fall, B.N. Diagana and LS. Holtzman, "Private Cereals Traders in
Southeastern Senegal and the APS: Observations on Opinions, Prospects and

Implementation Issues", Outline Paper prepared for Seminar and Discussion with
USAID/Dakar, December 8th, 1987, 10p.

Goetz, SJ. and LS. Holtzman with A. Dieng, ”Crop Mixes, Agicultural Inputs, and
Participation with Parastatal Organizations: Reports from Farmers in South-
eastern Senegal", ISRA/BAME Publication Series, May 1987, 30p.

Goetz, SJ. and LS. Holtzman, ”Input Acquisition and Crop Mix Changes in 1987: A
Preliminary Report for Southeastern Senegal”, ISRA/BAME Publication Series,
March 1988.

Gordon, D.V. "A Revenue-Function Approach to the Measurement of Output-Substitu-
tion Possibilities in Agiculture.” Jml. Eue. Econ. Stat. 7(4),1989:483-7.

Government of Senegal (GOS), Ministere de Developpement Rurale, Direction
d’Agiculture, ”Preliminary Crop Area, Yield and Production Estimates, 1986",
mimeo (draft document), 1986.

Haessel, W. "The Price Response of Home Consumption and Marketed Surplus of Food
Grains." Amer. J, Ag; Been. 57(1975):111-15.

Haines, P.S., D.K. Guilkey and BM. Popkin. "Modeling Food Consumption Decisions as

a Two-Step Process." W. 70(1988):543-52.
Hammer, J .S. ”Subsistence First’: Farm Allocation Decisions in Senegal.” J. Dev. Econ.,
23(1985):355-69.

Havard, M. ”Les Conclusions des Experimentations (1950-85) sur les Semis en Culture
Attelée des Principales Especes Cultivées', Dakar: ISRA, Département Systemes,
Document de Travail. 1986a.

Havard, M. 'L’Apport de la Recherche en Machinisme a la Mecanisation de l’Agicul-
ture au Sénégal”. Dakar; ISRA, Document provisoire, August 1986b (mimeo).

Heiner, RA. "The Origin of Predictable Behavior." W. 73, 1983:560-95.

247

Heiner, R.A. ”Origin of Predictable Behavior: Further Modeling and Applications."
W. 75,1985:391-6.

Holtzman, J .S. ”An Institutional Critique of mm from an African
Perspective", W 27(Winter l987):61-74.

Holtzman, J.S. ”Coarse Grain Processing in Senegal: Issues, Constraints, and Policy and
Progam Options.” Washington, DC: Abt Associates, May 1989.

Jolly, C., M. Kamuanga, S. Sall et J .L. Posner, "Situation Céréaliere en Milieu Paysan en
Basse Casamance: Résultats d’une Enquete de Terrain", ISRA, Département
Systemes et Transfert, Document de Travail, No.85-3, Dakar (January 1985).

J unankar, P.N. "The Response of Farmers to Price Incentives: The Use and Misuse of
Profit Functions.” ,lnnl. Dev, Slug. 25(2),1989:169-182.

Kelly, V. ”Factors Affecting the Demand for Fertilizer in Senegal’s Peanut Basin",
unpubl. PhD. Dissertation, Michigan State University, East Lansing, Michigan,
1988a.

Kelly, V. "Farmers’ Demand for Fertilizer in the Context of Senegal’s New Agicultural
Policy: A Study of Factors Influencing Farmers’ Fertilizer Purchasing Decisions",
MSU International Development Papers Reprint, No. 19, East Lansing,
Michigan, 1988b.

Kelly, V., "Acquisition and Use of Agicultural Inputs in the Context of Senegal’s New
Agicultural Policy: The Implications of Farmers’ Attitudes and Input Purchasing
Behavior for the Design of Agicultural Policy and Research Progams." MSU
International Development Papers Reprint, No. 18, East Lansing, Michigan,
1988c.

Kelly, V. and M. Gaye. "A Report on Input Acquisition and Use in the Departments of
Gossas and Nioro", Dakar: ISRA/BAME, prepared for USAID/Senegal, 1986
(mimeo).

Kim, M. and U. Ben-Zion, "The Structure of Technology in a Multioutput Branch
Banking Firm," J rnl. Bus. Econ. Stat., Vol. 7., No. 4, 1989:489-96.

Krishna, R. ”A Note on the Elasticity of the Marketable Surplus of a Subsistence Crop."
lnglia an 1. Ag}; Eco o.n 17(1962): 79- 82.

Kuroda, Y. ”The Output Bias of Technological Change 1n Postwar Japanese Agiculture,"

MW 70(1988) 663 73

Landais, E. "Elements pour la Preparation du Progamme de Recherches sur les
Systemes de Production et le Transfert de Technologies en Milieu Rural”, ISRA,
Departement Systémes et Transfert, Document de Travail, No.85-3, Dakar
(January, 1985).

248

Lambert, A. avec M. Diouf, "Espaces Réseaux Marchands au Sénégal - Les Echanges
Céréaliers avec la Gambie et la Mauretanie," Club du Sahel (OCDE), Ministere
de la Cooperation (Paris), CILSS (Ouagadougou), Mars 1989.

Lau, L., W.-L. Lin and P. Yotopoulos. "The Linear Logarithmic Expenditure System: An
Application to Consumption Leisure Choice.” Eegnennenjg, 46(1978):843-68.

Layard, P.R.G. and AA. Walters, WW McGraw-Hill Book Company,
New York, 1978.

Lee, H. and R.G. Chambers. "Expenditure Constraints and Profit Maximization in US.

Agiculture.” W. 68(4), 1986:857-65.

Lele, UJ. "Managing Agicultural Development in Africa: Lessons of Experience for
Government and Aid Donors.” Einenee Q Devel. March 1989, p.45-48.

Lele, U.J., N. van de Walle and M. Gbetibouo, ”Cotton in Africa: An Analysis of
Differences in Performance." World Bank, MADIA Discussion Paper 7, 1989.

Lombard, J. "Systeme de Production et Autosuffisance Céréaliere en Pays Serer

(Sénégal)." Qahiegs glee giences Hnrnninee, 23(3-4)1987:471-82.
Maddala, G.S. ' ' e e ta ual't 'v Va ' ' t ' . Econ. Soc.

Monog. No. 3, Cambridge University Press, Cambridge, 1988.

Martin, F. "Food Security and Comparative Advantage in Senegal: A Micro-Macro
Approach”, unpubl. PhD. Dissertation, Michigan State University, East Lansing,
Michigan, 1988.

McDonald, J. and P. Moffit. "The Uses of Tobit Analysis." Rev con. a Stat.
62(1980):318-21.

MDR/ Government of Senegal, ”La Nouvelle Politique Agicole,” Dakar, Mai, 1984.

MDR/Government of Senegal, "Etude du Secteur Agicole: Plan Céréalier", Dakar, Mai,
1986.

MDR/SODEFITEX-SONED, "Projet de Developpement Rural au Sénégal Oriental et
en Haute Casamance", 6 Volumes et 1 Atlas, Afrique-Courtoy Bureau d’Etudes
Associés, June 1980.

Monnier, J. ”Le Travail dans L’Exploitation Agicole Sénégalaise: L’Incidence de la
Division Sociale du Travail sure la Combinaison des F acteurs de Production et
sur la Productivité Globale du Travail en Pays Wolof, Sine-Saloum.” Centre
National de Recherches Agonomiques (CNRA) de Bambey, Institut de
Recherches Agonomiques Tropicales et des Cultures Vivrieres (IRAT). April,
1974. Mimeo.

Morris, M.L., 1988, "Rice Marketing in the Senegal River Valley: Research Findings and
Policy Reform Options", MSU International Development Paper, No. 8, East
Lansing, Michigan.

249

Ndoye, O. "Diagiostic et la Gestion des Cooperatives des Unites Experimentales au Sud
du Sine Saloum de 1968 a 1979", ISRA, Secretariat d’Etat a la Recherche
Scientifique et Technique, March 1980.

Olson, M. W Harvard University Press, Massachusetts,
1971.

Ouedraogo, I. and O. Ndoye, 1988, "Guide de Collecte des Prix Agicoles en Afrique du
Sud du Sahara: Lecons Tirées du Sénégal", paper presented at the Workshop on
Research and Agicultural Food Policy, Dakar, Senegal, July 7-8th.

Panzar, J .C. and RD. Willig ”Economies of Scale in Multi-Output Production.” 9%
IIOLEEQO- XCI(3), 1977:481-93.

Panzar, LC. and RD. Willig "Economies of Scope." W. 7 1(2)1981:268-
72.

Poirier, DJ. "Partial Observability in Bivariate Probit Models." Jml. of Econemetrics,
12,1980:209-17.

Reardon, T., P. Matlon and CL. Delgado. "Coping with Food Insecurity at the
Household Level in Drought-Affected Areas of Burkina Faso." Werld Develop.
16(1988):1065-1074.

Reardon, T., T. 'I'hiombiano and C. L. Delgado, "L’Importance des Céréales Non
Traditionnelles dans la Consommation des Riches et des Pauvres a

Ouagadougou," W Mars-Avril, No.190,1989.

Renkow, M. "Household Inventories and Marketed Surplus in Semisubsistence Agicul-
ture." Paper presented at the 1988 Amer. Agic. Econ. Assoc. meetings,
Knoxville, Tennessee, July 31-August 3. (mimeo)

Rosett, R.N. "A Statistical Model of Friction in Economics." Econometrica 27(1959):
263-7.

Ross, C.G. ”Grain Demand and Consumer Preference in Senegal," Eoog Eolig,
5(1980):273-281

Schmid, A.A. "Analytical Institutional Economics: Challenging Problems 1n the
Economics of Resources for a New Environment. A J c c
54, 1972: 893-901.

Schmid, AA. and L. Robison. "Interpersonal Relationships and Preferences: Evidences
and Implications." MSU Staff Paper No. 88-21. Dept. Agic. Economics, Michigan
State University.

Schmidt, P. Econometg’ee. Marcel Dekker, Inc., New York, Statistics Textbooks and
Monogaphs, Vol 18. 1976.

250

Sen, A., 1987, "Poverty and Famines", pp. 198-204 in Minerals. Ingegrnjingfinnnly,
Winn, by J. P. Gittinger et al., EDI, John Hopkins/World
Bank, Baltimore. (orig. 1982).

Shaffer, J. D. "Food System Organization and Performance: Toward a Conceptual Frame-
work" W 62(2)1930-310°8

Shaffer, J. D., M. T. Weber, H. M. Riley and J. Staatz. "Desigiing Market Systems to
Promote Development 1n Thirdi World Countries. In ICRISAT, Agn'nnanl
t S - d Proceeding of the International Workshop
held at ICRISAT Center, India, 24-28 October, 1983.

Singh, I, L. Squire and J. Strauss.

MW: Mansions.
Annhnnnnnunifinhny, John Hopkins University Press (World Bank), Baltimore,
1986.

SODEFITEX, "Rapport d’Activités - Campagne Agicole 1985-86", Ministere du
Developpement Rural, Société de Developpement des Fibres Textiles, Dakar,
1986.

SONED-SODETEG, "Etude d’un Plan de Developpement Regional Integé du Sénégal
Oriental". Dakar, July 1977.

Staatz, J. M. "The Cooperative as a Coalition: A Game Theoretic Approach." Amei._J_m_l.
W.65(5)1983:1084-9.

Staatz, J. M., J. Dioné and N. N. Dembélé. "Cereals Market Liberalization 1n Mali." _nr___Wld
Dflelnn. 17(5)1989z703-18.

Stiglitz, J.E. "Markets, Market Failures, and Development." Amer. Econ. Rev. 79(2),
1989: 197-203.

Stomal- -Weigel, B. "L’Evolution Récente de Comparée des Systemes de Production Serer
et Wolof dans Deux Villages du Vieux Bassin Arachidier (Sénégal)." Qaniers des

Seience sflumaines, 24(1)1988:17-33.

Strauss, J. "Marketed Surpluses of Agicultural Households in Sierra Leone". AIDELL
Mn. 66(1984):321-31. .

Strauss, J. "Does Better Nutrition Raise Farm Productivity?" ,lml. Pol. Egon,
94(2)1986:297-320.

Sukhatme, P. V. "Incidence of Protein Deficiency 1n Relation to Different Diets 1n India,"

E[il. ,lnnl, Null}, 24, 1970: 447-87.

Thurow, LC. 3 e o u e ts: tat ics. Vintage Books, Random
House, New York. 1984.

Timmer, C.P., W.P. Falcon and SR. Pearson. Ennd Eolig Analysin. Baltimore; The John
Hopkins University Press (World Bank), 1983.

251

USAID/APS, "Agicultural Project Support" Progam Document, Dakar, Senegal, 1987,
m1meo.

USAID/Senegal, "Progam Assistance Approval Document: Agicultural Assistance
Sahel Development Fund", No. 685-0249, Dakar, Senegal, 1983.

USAID/Data Base, "Agicultural Data Base for Senegal," USAID/ Dakar, Senegal,
(Compiled by M. Keita, various dates), 1987

Varian, H.R. MignennnnminAnnlynin. W.W. Norton & Company, New York, 2nd ed.
1984.

von Braun, J. "Effects of Technological Change in Agiculture on Food Consumption and
Nutrition: Rice in a West African Setting," Worln Devel., 16(1988):IO83-1098.

von Braun, J. and E. Kennedy. "Commercialization of Subsistence Agiculture: Income
and Nutritional Effects in Developing Countries", Working Papers on
Commercialization of Agiculture and Nutrition, No. 1, Washington, DC, IFPRI,
April 1986.

von Pirschke, J.D., D.W. Adams and G. Donald. Enignl Financial Mailgels in Developing
mm: fljneir Use nnn Abuse, John Hopkins University Press (Economic
Development Institute, World Bank), Baltimore, 1983.

Weber, M.T., J.M. Staatz, J.S. Holtzman, E.W. Crawford and RH. Bernsten. "Informing
Food Security Decisions in Africa: Empiricial Analysis and Policy Dialogue."
MQLJMEeen. 70(1988)=1044-52-

Williamson, O.E. Economic Qrgang' ation: Eimis, Markets and Poliny Contiol, Brighton,
Wheatsheaf, 1986.

Willig, R.D. "Multiproduct Technology and Market Structure." Amen, Econ. Rev. 69(2),
1979:346-51.

World Bank, "Price Prospects for Major Primary Commodities", Washington, DC,
November 1988, (mimeo).

Yotopoulos, RA. and LJ. Lau. "On Modeling the Agicultural Sector in Developing
Countries," Jml. Dev. Econ. 1(2)1974:105-27.

MICHIGAN STATE unxv. LIBRnRIE
lllWWWWW“lllllllllllllllllllHlHlWl
31293007695152