. h v
. “EL...
‘

-9! b mun w.

.. . é - .

. 2.. AW. 3a"

, .Lotnlu,

- !\ I ’
’1

t. .

Una: HIV
{0" . «new.

I. it...
til: .

I c In. lid-{.3
. ht. .l.a.‘u.|..‘kl|
. ...|A .v.\k

=

1.1.»:
3.1

in“

wry»);
fi

2 . .5; I.
, .4“..\Cp
u -(.JI\‘0W .1:
1.1..

“pituitl‘l

b:

I. ‘lbhucttv . .
‘19.! .. .a \.

 

«1
.910

. . Al,
I29...) ‘11 . r \o
.p it. 5.6:}

(r .. Vra
f 1.1‘} :‘l

(‘2...
.

1‘.“ I. ..|.

‘Vtil..¢\lv

. is
inhuman»

. {yarnfivavclug .r. :3-

in! ..
:-
(1.:
.v ‘
I l .0! 1....
15.3.“..‘n.

fiéfiw
..\.
e‘... . .

. as

5!). I
:1... .
9?»!

. «Nahum?
Y..|¢\‘..‘A ‘ I 'I?
.6. .xn‘ufinffilh

. 1.1;. .Lusn
.. ‘ 2%.... mu.
15.3% Hi}. 1“.
fiafiuwmauammwflw

Lu. i

‘3: u

. , ,1o~1>.| “'4'...
4 . ru‘..!.-..L.W!unnuvtlxsu.
‘ 2:11;...9...)
. ‘l.~‘i'\l 5". -.
Kaynnl‘l
his:
.11.:

5‘. 9)... ‘
0‘.‘v|.:lt
. 1.5.771}. .
. 1;, \..v . 1"»!
A.» 3....Vnzlll
0:1. .1_.|l.‘

$J§J§flfr . nvl‘vnu u
l ..1 .nll‘|vl!nu .Y 'ltslv.\.

 

THEBlS

                                                                                                                           

 

 

\ 302050 9687 E "are n nu a
’3 ,. W) ) 1“...
_. t" .J '5‘.
Luis-a Why”?-
up sets-w .

 

 

This is to certify that the
dissertation entitled
ESSAYS ON THE ECONOMICS OF SOIL NUTRIENT REPLENISHMENT IN

ECOLOGICALLY FRAGILE REGIONS OF SUBSAHARAN AFRICA:
EVIDENCE FROM SENEGAL

presented by

Bocar Nene Diagana

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Agricultural Economics

6%

Major professor

 

 

 

 

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

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINE return on or before date due.
MAY BE RECAUED with earlier due date if requested.

 

DATE DUE DATE DUE! DATE DUE

 

U9 '2’ 3 2.002

 

1233330037 tum

 

 

 

 

 

 

 

 

 

 

 

 

 

11m www.m-pu

ESSAYS ON THE ECONOMICS OF SOIL NUTRIENT REPLENISHMENT IN
ECOLOGICALLY FRAGILE REGIONS OF SUBSAHARAN AFRICA:
EVIDENCE FROM SENEGAL

By

Bocar Nene Diagana

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

1999

 

ABSTRACT

ESSAYS ON THE ECONOMICS OF SOIL NUTRIENT REPLENISHMENT IN
ECOLOGICALLY FRAGILE REGIONS OF SUBSAHARAN AFRICA:
EVIDENCE FROM SENEGAL

By

Bocar Nene Diagana

Soil fertility decline has been said to constitute a major cause of low agricultural
productivity and a threat to food security in SubSaharan Africa, especially in its ecologically
fragile regions. Already nutrient-poor soils, subjected to continuous cropping, wind and water
erosion, are mined of their nutrients by farming practices that include very low use of mineral
and organic fertilizer. Reversing nutrient depletion of soils by replenishing macronutrient
pools requires policy measures that provide incentives and improve farmers’ capacity to make
the necessary short- and long-term investments in land productivity and quality maintenance.

Price policy, credit and capital input distribution are commonly used policy
instruments to influence farmers’ long-term production choices of activities and technologies.
The effects of these policies on soil fertility are not clear and direct; they are mediated
through farmers’ responses which can take different paths (extensify or intensify crop
production in a sustainable or unsustainable way). These paths and their conditioners are not

well understood, though they are crucial to determining the fate of these policies.

 

This research uses the semiarid area of the Senegalese Peanut Basin to empirically
explore the long-term production paths followed by farmers in response to selected policies
and their subsequent impacts on soil nutrient pools. It is organized in three interrelated essays.

Essay one uses a theoretical dynamic farm household model and shows that
intertemporal tradeofl‘s (current versus fixture output due to soil nutrient replenishment or
mining) and time preferences, i.e., discount rates afi‘ect farmers’ optimal input (especially
fertilizer) decisions.

The second essay uses a biophysical crop simulation model to predict the plot-level
production and soil nutrient impacts of a set of cropping activities (millet and peanut rotation)
and technological (fertilizer-based versus others) choices. Results confirm that in the Peanut
Basin of Senegal fertilizer-based cropping practices lead to more millet and peanut crop
output, contribute more to replenishing soil nutrients, and are financially more attractive in
the long term than practices that do not use fertilizer.

The third essay uses a bioeconomic model that integrates simulated plot-level
biophysical outcomes with current and fixture input/output prices, and farm household
resources and objectives to predict the effects of selected policy measures on the farm
household’s optimal crop and technology choices, and their implications in terms of soil
nutrient replenishment. Optimal cropping practices on millet and peanut suggested by a multi-
period (10-year) linear programming model are those that use fertilizer and lead to
replenishing the plant-available soil nutrient pools. However, overcoming initial financial
constraints is key to launching the process of intensifying crop production through increased

fertilizer use.

DEDICATION

This dissertation is dedicated to my father Abdoulaye who did not live long enough to see
me become who I am today, but was able to sow very early in me the seeds of hard work,

family spirit and faith in God.

iv

 

ACKNOWLEDGIVIENTS

First of all, all my thanks go to ALLAH The Almighty for having blessed me with the
opportunity and the capacity in all dimensions to engage in and successfully complete this
doctoral program!
Secondly, my sincere thanks to:
- USAID for having fiJnded in its entirety my academic program under the Food
Security Project II of the Dept of Agricultural Economics of Michigan State
University;
-. my advisor Dr. T. Reardon for constantly believing in me, challenging me to always
do a bit more while backing it up with an unconditional all-around support;
- Drs. E. Crawford, J. Ritchie, J. Strauss and S. Swinton for serving in my committee;
- Drs. S. Daroub and A. Gerakis, along with B. Baer of the Dept of Plant and Soil
Sciences at Michigan State University whose tremendous assistance was instrumental
in sharpening my (biophysical) modeling skills;
- Dr. S. Harsh for his review and suggestions on my linear programming model;
- Drs. V. Kelly and A. Fall, M. Sene and O. Ndoye for their help in data collection.
Thirdly, my deepest and heartfelt gratitude to:
- my wife Thioro and my daughters Nabou and Rokhaya who, silently and with

dignity, endured my being away for almost five years and provided me with a

 

relentless support that helped me go through the challenges, frustrations and
loneliness of this journey;
- to my mom, dad, my brothers and sisters, nephews and nieces back in Thies, Senegal

‘ for their prayers and encouragement during every walk of this endeavor.

TABLE OF CONTENTS

List of Tables ........................................................ xii
List of Figures ...................................................... xiv
List of Symbols or Abbreviations ........................................ xv
INTRODUCTION

1. Soil Fertility Decline, a Major Cause of Low Agricultural Productivity in Sub-

Saharan Afiica ................................................. 1
2. The Productivity Growth Challenge and its Soil Fertility Implications ....... 3
3. Soil Replenishment Alternatives ................................... 4

3.1 Technological solutions .................................. 4

3.2 Policy alternatives ....................................... 6
4. Problem, Gap and Research Objectives ............................. 6
5. Data/Context/Research Questions ................................. 9
6. Research Methods ............................................ 10
7. Organization of Thesis ......................................... 10
References .................................................... 12
Appendix A1 .................................................. 17

Presentation of the Agricultural and Phosphate Program in Senegal . . . 17

vii

 

ESSAY I: MICROECONOMICS OF SOIL NUTRIENT REPLENISHMENT: A
THEORETICAL DYNAMIC MODEL OF FARM PRODUCTION

BEHAVIOR
1. Problem Statement ............................................ l9
2. Profitability of Fertilizer Use in SSA ............................... 20
2.1 Mixed evidence ........................................ 20
2.2. Dynamic dimensions ................................... 21
3. Objective of Essay ............................................ 23
4. Dynamic Bioeconomic Model of Farm Household Behavior ............ 24
5. Solving the Model .................... i ........................ 27
6. Discussion of Theoretical Results ................................. 28
6.1 Interpreting the first-order conditions ....................... 28
6.2 Predicting the impacts of selected policy measures ............. 30
7. Conclusions ................................................. 31
References .................................................... 32

ESSAY II: LAND QUALITY AND PRODUCTIVITY IMPACTS OF SELECTED
CROPPING PRACTICES: AN APPLICATION OF BIOPHYSICAL
MODELING TO THE SAI-IELIAN CONTEXT OF SENEGAL

1. Introduction ................................................. 35
2. Research Objectives and Questions ............................... 38
3. Cropping Systems in the Senegalese Peanut Basin .................... 39
4. Materials and Methods ......................................... 43

viii

4.1 Materials ............................................ 43

4.2 Simulation Experiments ................................. 45

4.3 Simulation Methods .................................... 47

4.3.1 Description of the crop growth simulation model ....... 47

4.3.2 Model modification and testing ..................... 48

4.4 Data Analysis Methods .................................. 50

5. Results and Discussion ......................................... 51
5.1. Model Validation Results ................................ 51

5.2. Simulation Results ..................................... 54

5.2.1. Long term biophysical outcomes of cropping practices . . 54
5.2.1.1 Yield effects ........................... 54
5.2.1.2. Soil nutrient effects ...................... 56

5.2.2. Stochastic dominance analysis of cropping practices . . . . 59

6. Conclusions ................................................. 61
References .................................................... 63
Appendix A2 .................................................. 68

Table A.2.a: Agronomic characteristics of millet production in the Peanut
Basin (PB) .............................................. 68

Table A.2.b: Agronomic characteristics of peanut production in the Peanut
Basin (PB) .............................................. 68

Appendix A3 .................................................. 69

ix

 

Table A3.1.a: Summary of simulated water balance component results in
Center Peanut Basin for no-fertilizer treatment ................... 69

Table A3.1.b: Summary of simulated water balance component results in
Center Peanut Basin for semi-intensive fertilizer treatment .......... 70

Table A3.2.a: Summary of simulated water balance component results in
South Peanut Basin for no-fertilizer treatment .................... 71

Table A3.2.b: Summary of simulated water balance component results in
South Peanut Basin for semi-intensive fertilizer treatment ........... 72

ESSAY III: MICROECONOMICS OF SOIL NUTRIENT REPLENISHMENT 1N SUB-
SAHARAN AFRICA: EVIDENCE FROM SENEGAL AND PROSPECTS

FOR A SUSTAINABLE FARM INTENSIFICATION
1. Introduction ................................................. 73
2. Cropping Practices in the Senegalese Peanut Basin .................... 77
2.1 Context ............................................. 77
2.2 Cropping practices ..................................... 79
2.3 Simulated biophysical outcomes ........................... 80
3. Method and Data ............................................. 83
3.1 Method .............................................. 83
3.2 Data ................................................ 85
4. Empirical Farm Household Model ................................ 86
4.1 Activities ............................................ 88
4.2. Constraints .......................................... 88

4.3. Right-hand side (RHS) values ............................ 90

4.4. Other assumptions and scenarios .......................... 91
5. Results ..................................................... 93
6. Conclusions ................................................. 96
References .................................................... 99

 

Table 1. 1a:

Table 1.1b:

Table 1.2a:

Table 1.2a:

Table 1.3:

Table 1.4a:

Table 1.4b:

Table 3.1:

Table 3 .2:

Table 3.3:

Table 3.4:

LIST OF TABLES

Weather Characteristics, Center Peanut Basin, Senegal ............. 44
Weather Characteristics, South Peanut Basin, Senegal ............. 44
Soil Characteristics, Center Peanut Basin, Senegal (site: Colobane) . . . 46
Soil Characteristics, South Peanut Basin, Senegal (site: Dioly) ....... 46

Simulated long-term yield effects of selected millet-peanut cropping practices
in the Senegalese Peanut Basin: 1977-96 (kg/ha/year) .............. 55

Simulated average long-term effects of millet-peanut cropping practices on
key soil nutrient pools in the Center Peanut Basin, Senegal: 1977-96 (in
kg/ha/year) .............................................. 57

Simulated average long-term effects of millet-peanut cropping practices on
key soil nutrient pools in the South Peanut Basin, Senegal: 1977-96 (in
kg/ha/year) .............................................. 58

Simulated long-term average yield and soil nutrient effects of selected millet-
peanut cropping practices in the Senegalese Peanut Basin: 1977-96
(kg/ha/year) ............................................. 82

Selected characteristics of the typical farm household in the Senegalese Peanut

Basin .................................................. 85
Simulated mean millet and peanut yields (kg/ha) by state of nature . . . . 90

Summary of the LP farm model (one period only) ................. 92

xii

 

Table 3.5: Optimal long-term cropping plan (in ha/year) for the typical farm household
in the Senegalese Peanut Basin under difl‘erent scenarios ........... 94

xiii

 

Figure 1.0:

Figure 1.1:

Figure 1.2:

Figure 1.3:

Figure 1.4:

Figure 1.5:

Figure 1.6:

Figure 1.7:

LIST OF FIGURES

Organization of research analysis ............................. 8
Total annual rainfall in Senegal ............................... 41
Average millet and peanut yields in Senegal ..................... 41

Total inorganic fertilizer (all formulas) consumption in Senegal (in metric
tons) ................................................... 42

Observed vs simulated yields (kg/ha) of a hectare of millet-peanut rotation in
the Senegalese Peanut Basin: 1977-82 ......................... 52

Cumulative sum of mean millet-peanut yields in the observed and simulated
experiments at three levels of NPK fertilization in the Center Peanut Basin,
Senegal ................................................ 52

Cumulative probability distribution curve of gross margins over variable costs
of a ha of millet-peanut rotation grown under selected management practices
in Center Peanut Basin, Senegal: 1977-96 ....................... 60

Cumulative probability distribution curve of gross margins over variable costs
of a ha of millet-peanut rotation grown under selected management practices
in South Peanut Basin, Senegal: 1977-96 ....................... 60

xiv

CEC:

CIMMYT:

DSSAT:
FAO:
GRS:
IBSNAT:
ICS:
ICRAF:

IFDC:

ISRA:

LP:

Mg:
MOTAD:

MSU:

.2.

LIST OF SYMBOLS AND ABBREVIATIONS

Cation Exchange Capacity

Centro Intemacional de Mejoramiento de Mais Y Trigo
Decision Support System for Agrotechnology Transfer
Food and Agriculture Organization of the United Nations
Groupe de Reflexion Strategique

International Benchmark Sites Network for Agrotechnology Transfer
Industries Chimiques du Senegal

International Center for Research on AgroForestry
International Fertilizer Development Center

International Food Policy Research Institute

Institut Senegalais de Recherches Agricoles

Linear Programming

Ministry of Agriculture

Magnesium I

Minimization Of Total Absolute Deviation

Michigan State University

Nitrogen

Nonfarm

Phosphorus

 

PB:

PP:

PRISAS:

SD:

SSA:

SSPT:

UNDP:

UNECA:

Zn:

Peanut Basin

Phosphate Program

Projet de Recherche Institutionnelle sur la Securite Alimentaire au Sahel
Sulfur

Stochastic Dominance

Sub-Saharan Africa

Societe Senegalaise des Phosphates de Thies

United Nations Development Program

United Nations Economic Commission for Africa

World Bank

Zinc

 

INTRODUCTION

1. Soil Fertility Decline, a Major Cause of Low Agricultural Productivity in Sub-
Saharan Africa:

Agricultural productivity and food security in Sub-Saharan Africa (SSA) are being
seriously threatened by the steady decline in soil fertility, defined as “a net decrease in
available nutrients and organic matter in the soil”1 (Scherr, 1999), and caused by the
continued mining of soil nutrients by farmers seeking to increase output. Declining soil
fertility jeopardizes the sustainability of farming systems in regions of SSA, especially those
in semi-arid West Afiica that are ecologically fragile. Cultivated soils in these areas are poorly
endowed in macronutrients (N, P, 8, Mg, Zn), heavily leached, acid and have low soil
organic matter (Wong et al., 1991).They have also been subjected to continuous cropping,
wind and water erosion. These characteristics determine their fertility status, hence their
agricultural potential.

Brady (1990) estimated only 12% of African soils to be “moderately fertile, well-
drained soils”, compared to 33% in Asia. Highly variable and declining rainfall patterns
observed since the 19705 compound the ecological fragility of the arid and semi-arid regions

which account for half of the cultivable land in SSA (Marter and Gordon, 1996). It is also

 

1 This view of soil fertility effects can be contrasted with alternative perspectives from
other disciplines:
- an agronomic one which defines soil fertility as “the capacity of soils to create
more food of high quality; ...food is fabricated soil fertility” (Sheldon, 1987) or
- a soil scientist one for which it refers to the “capacity of a soil to supply essential
elements (nutrients) for plant growth without a toxic concentration of any
element” (Foth, 1990).

 

estimated that 65% of SSA's agricultural land is degraded because of water and soil erosion,
chemical and physical degradation (Oldeman et aI. , 1991; Scherr, 1999). Research has shown
that soil nutrient depletion resulting from soil mining or the practice of growing crops with
insuficient replacement of macro-nutrients removed from the soil is an important problem in
low income countries (Bishop and Allen, 1989; Stocking, 1987; Stoorvogel and Smaling,
1990), and a fundamental biophysical constraint to steady growth of food production
(Donovan and Casey, 1998). On a per ha basis, 22 kg N, 2.5 kg P and 15 kg K are being lost
annually as a result of long-term cropping with little or no external nutrient inputs and
returned crop residues (Stoorvogel and Smaling, 1990; Smaling er al., 1993;Weight and
Kelly, 1998).

Particularly serious is the phosphorus (P) deficiency that affects 80% of SSA’s soils.
Studies by IFDC and others have firmly established that P is the most limiting nutrient in soils
in semi arid West Africa (Bationo and Mokwunye, 1991; Bationo and Vlek, 1997). This
deficiency not only affects plant growth and crop quality, but it also constrains response by
crops to other nutrients (Gemer and Mokwunye, 1995; WB/IFDC/ICRAF, 1994; Jones et a1. ,
1991; Brady and Wei], 1996). Consequently, increasing the supply of available soil P is
essential for productivity growth (increasing crop yields) and environmental quality (stopping
or slowing land degradation). Since the P content of crop residues and manures does not
usually cover crop requirements, P fertilizer inputs are almost always necessary to correct P

deficiency (Bremen, 1990; McIntire and Powell, 1995).

2. The Productivity Growth Challenge and its Soil Fertility Implications:

Partly as a result of these adverse agroecological conditions in SSA, agricultural
production grew annually at less than 2% between 1965 and 1980, and at around 1.4% during
the 19908 (UNDP/UNECA Report, 1997), well under the rapid pace of demographic growth
at around 3% per year. To meet soaring food and fiber needs from a fast growing population,
it’s been argued that agricultural production should grow at an estimated rate of 4% per
annum. This would then require an annual increase of 1.5% for labor productivity and of 3%
for land productivity (Delgado et al., 1987; Cleaver and Schreiber, 1992; Larson and
Frisvold, 1996).

Several production paths can be theoretically envisioned to meet this serious
agricultural productivity growth challenge. First, extensification by expanding on to new and
marginal lands offers limited potential as the population pressure on the available agricultural
land has prompted a nearing of the land frontier, making it much more difficult or even
impossible to increase production on the existing but degraded farmlands. Worse,
extensification is likely to put further pressure on forested areas and resources, leading to
more land degradation and deforestation (Marter and Gordon, 1996).

Second is the intensification of agricultural production by using more productivity-
enhancing inputs per unit of land area (improved seeds, chemical inputs, labor). But,
intensification paths can be of different types: sustainable and unsustainable (Reardon et al.,
1997; 1999). One major difference between them depends upon whether they result in
negative or positive soil fertility and productivity impacts. Whatever the production path,

there is a growing consensus that the appropriate one capable of meeting the productivity

challenge must be sustainable, i.e with a real potential to reverse the declining crop yield
trends, to ensure a concomitant and appropriate replenishment of soils while still being

profitable

3. Soil Replenishment Alternatives:
3.1 Technological solutions:

To reduce net soil nutrient losses, hence soil fertility decline, several options are
available to farmers: crop rotation, fallows or fertilization, or a combination of them. Rotating
or sequencing crops, for example nitrogen-fixing legumes followed by nitrogen-demanding
cereals on a given piece of land, allows a smoothing of macronutrient consumption across
crops. Higher crop yields than those from monocropping could follow (Bationo and Lompo,
1999), in addition to reduced soil erosion, less negative environmental extemalities, better soil
fertility and less need for commercial fertilizer (Gebremedhin and Schwab, 1998). Where land
is abundant relatively to labor, regenerative long-term fallows or shorter improved fallow
techniques are also used to replenish soil nutrients. But these fallow strategies have been
progressively abandoned or reduced under the pressure of high population densities (Dalton,
1996).

In the mean time, the utilization of productivity-enhancing inputs and technologies is
very low. A key intensifying input such as fertilizer is only sparsely used. Average fertilizer
use has been estimated in SSA at less than 15 kg per hectare as of 1994/95, compared to
more than 200 kg in East Asia, 125 kg for Asia as a whole and 65 kg in Latin America

(UNDP/UNECA Report, 1997). Annual growth in fertilizer consumption per ha has been

declining in the 903: .3 % between 1990 and 1993, and negative between 1993 and 1995. Use
of manure is constrained by availability. McIntire and Powell (1995) underscore the enormity
of the required pasture areas to produce enough manure to maintain soil fertility in the
absence of mineral fertilizer whereas Williams et al. (1995) claim that, with present intensity
of land use in semi-arid West African countries, manure alone would not increase crop yields
in a sustainable manner. Competing livestock feed, fuel and construction uses of crop
residues constrain their availability for incorporation in soils. Consequently, soil organic
matter drops, lowering crop yields or soil productivity (Bationo and Vlek, 1997).

Given the feasibility limits of the technological options presented above, increased
inorganic fertilizer is the remaining option (Mudahar, 1986), along with soil organic matter
improvement. As concluded by Padwick (1983) in his review of 50-year soil fertility studies
in tropical Africa, “inorganic fertilizers are an essential part of any system aimed at
maintaining good yields over large areas in the absence of sufficient organic manures”. To
“jump-start” the process of soil replenishment, use of inorganic fertilizer must be increased
first from its current 10-15 kg/ha levels to at least 30-50 kg/ha 2(Larson and Frisvold, 1996;
Weight and Kelly, 1998), helping thereby to reduce soil organic matter loss, sustain crop
production, raise agricultural productivity, improve food security and preserve the natural

resource base, all features of a sustainable intensification of agricultural production.

 

2 While overuse of fertilizers has lead to environmental problems in other parts of the
globe, applying this level of inorganic fertilizer should not cause similar problems in SSA.

5

3.2 Policy alternatives:

For policymakers, the challenge is to put in place policies that provide suitable
incentives to farmers and improve their capacity to increase factor productivity while
maintaining appropriate levels of the physical resource base (Kruseman et al., 1993). Under
market-oriented reforms, product and factor prices, financial and physical capital transfers are
commonly used instruments by policymakers to influence the behavior paths described
earlier. One objective of these policies which has recently received a lot of attention is the
World Bank-led soil recapitalization programs in SSA (WB/FAO, 1996). These programs
treat the soil as a natural capital asset3 in whose maintenance or fructification both farmers
and society have an interest. This natural capital resource provides service flows positively
(negatively) influenced by soil replenishment (mining). Thus, where soil fertility is declining,
reversing it by replenishing the soil should be interpreted as a process of “recapitalizing” the

soil asset (de Alwis, 1995; Sanchez et al., 1995).

4. Problem, Gap and Research Objectives:

Achieving this long term objective depends a great deal upon the prevailing physical
and socioeconomic environment, i.e the links to agricultural input and output market
conditions (Debrah and Koster, 1999) shaped by policy measures that influence farmers’
incentives and capacity to engage in soil recapitalization. These policies, however, do not

have clear, direct and linear effects on soil fertility outcomes. The reality is that these effects

 

3 Along with water, atmosphere, forests, fish, wildlife and wetlands, soil is one of the
environmental assets that make up the stock of natural capital (Sanchez et al., 1997).

6

are mediated by farm producers’ behavior in response to policy shocks. Their response can
follow different paths. These paths and their conditioners (activity and technology choice,
agroclimatic context) are not well understood, but are crucial to determining the performance
of the applied policies. Moreover, the evidence is mixed on the farm profitability of these
conditioners.

Thus, the challenge is for policy researchers to fill this empirical knowledge gap. The
main objective of this research is therefore to empirically explore these farm production
behavior paths and their conditioners using the Senegalese physical and economic context as
a study case.

From this overall research objective, there follow two specific empirical analyses
around which this work is organized (figure 1). The discussion of these analyses is preceded
by the presentation of a theoretical dynamic framework to understand intertemporal
considerations involved in optimal farm household cropping decisions to ensure soil nutrient
replenishment. Then, the first empirical analysis uses a biophysical model of crop production
to measure the production and soil fertility impacts at the plot level of a set of conditioners
available to farm producers, i.e an array of cropping activities and technological (fertilizer-
based versus others) choices under the agroclimatic conditions of our study area. A second
empirical analysis includes these measured plot-level outcomes along with input/output prices
in a linear programming model to predict the effects of selected policy measures on the farm
household optimal crop and technology choices and their implications in terms of soil fertility

outcomes. Particular attention is paid to the issue of the long term profitability of inorganic

 

aim—new £0382 Lo cosafieawho ”_ 2:me

 

 

 

 

    
   

v.52 ~02 -
«3:23 255:: zom-
3052.5 mew—no.8 3 3286:“ ES— -

30:83 gimp—U

 

 

 

  
  

”ea—m Eenomsom 58m 1:530

 

 

 

35.5 25.522

 

 

   
   

 

game
.362 eons—=55 9:0

3:33?
82:82 - L

 

22>

 

 

 

”—3662 2283: :55
wEEEEmoE c354

 

 

 

“whosflnmzu _

 

 

 

 

 

 

 

 

as? 28:58 :33? 2.83

 

 

 

 

 

 

 

 

fertilizer use at the farm level in SSA. By focusing on this still-debated fertilizer profitability
issue, we are hoping to add much needed evidence to the body of existing but conflicting
findings fi'om other works in the region (McIntire, 1986; Adesina et al. , 1988; Diagana et al. ,

1995; Sanchez et al.,1997; Coulibaly et a1. , 1998; etc).

5. Data/Context/Research Questions:

The Western Sahelian country of Senegal provides the study case for our research.
This application will offer interesting policy insights in that it is based on a set of common
price and capital transfer policy measures contained in the multi-year ‘ Agricultural Program’
launched in 1997/98 (Republic of Senegal/MA, 1996; see details in appendix A 1) and seeking
to recapitalize soils in a country which presents most of the agroecological and
socioeconomic features discussed earlier in SSA: a) smallholder rainfed agriculture, b)
ecologically fragile areas with variable and low rainfall and poor soils, c) rapid demographic
growth, d) unstable price and input distribution policies, e) low rural incomes and limited
cash availability, and f) low fertilizer use.

Data have been collected on socioeconomic (factor and product prices, household
characteristics) and biophysical variables (weather, soils, cultivars, yield and input use on
peanut and millet) in two zones of the Senegalese Peanut Basin. They will help answer the
following general research questions:

(i) Have fertilizer-using cropping practices become more profitable at the farm level

in the long run than others that do not use fertilizer under the ‘Agricultural Program’

context in Senegal?

(ii) If not, under what price conditions would fertilizer use be profitable at the farm
level?
(iii) What are the corresponding soil fertility impacts, i.e on soil macronutrient

balances, especially N and P?

6. Research Methods:

To meet our research objectives, we are taking a relatively recent and
pluridisciplinary-based methodological route that directly links biophysical soil-plant-weather
mutual interactions (physical realm of crop production) to the economic decision analysis
(behavioral side of crop production). Feeding economic analysis with inputs from other
disciplines and vice versa has led to the development of bioeconomic models in the
agricultural production literature in the last two decades (Oriade and Dillon, 1997; Roberts
and Swinton, 1996). Bioeconomic models, though complex because of the breadth of their
scope, are handily solved by today’s computer programs, lend themselves for multi-faceted
policy analySis, and offer great perspectives for bridging gaps between academic disciplines
and between scientists and policymakers (Ruben et al., 1998). Our work will use a crop
growth simulation model whose results are linked to a linear programming-based farm

household model to explore our general research questions.

7. Organization of Thesis:
This dissertation is organized in three related, but separately treated essays. The first

essay is a short attempt to construct a theoretical dynamic fann- household model that

10

analyses optimal crop management decision rules (for example how much fertilizer to use)
under a soil nutrient replenishment concern. The second essay centers on using a crop
simulation model to measure plot-level crop yield and soil fertility impacts of selected crop
rotation practices. The third essay uses these plot-level results to analyze the optimal farm
household production choices under various policy scenarios with a special focus on fertilizer

long term profitability.

ll

References

Adesina, A.A., P.C. Abbott and J .H. Sanders. 1988. Ex-ante Risk Programming Appraisal of
New Agricultural Technology: Experiment Station Fertilizer Recommendations in
Southern Niger. Agricultural Systems, 27:23-35.

Bationo, A and F. Lompo. 1999. Available Technologies for Combating Soil Nutrient Losses
in West Afiica. in Debrah, SK. and W. G. Koster (eds). Linking Soil Fertility
Management to Agricultural Input and Output Market Development: the key to
Sustainable Agriculture in West Afiica. IFDC-Afiica, Lome, Togo.

Bationo, A and P.L.G. Vlek. 1998. “The Role of Nitrogen Fertilizers Applied to Food Crops
in the Sudano-Sahelian Zone of West Africa” in G. Renard, A. Neef, K. Becker and
M. Von Oppem (eds) “Soil Fertility Management in West Afiican Land Systems”
Proceedings of the University of Hohenheim (Germany), ICRISAT Sahelian Center
and INRAN regional workshop, 3-8 March 1997 Niamey, Niger. Werkersheirn:
Margraf Verlag.

Bationo, A. and A.U Mokwunye. 1991. Alleviating Soil Fertility Constraints to Increased
Crop Production in West Afiica. The Experience in the Sahel. p.195-215. In A.U.
Mokwunye (ed.) Alleviating Soil Fertility Constraints to Increased Crop Production
in West Afiica. Kluwer Academic Publications, Dordrecht, the Netherlands.

Bishop, J. and J. Allen. 1989.0n-Site Costs of Soil Erosion in Mali. Environment Department
Working Paper No. 21, World Bank, Washington DC, November.

Brady, NC. 1990. The Nature and Properties of Soils. Tenth edition. New York: MacMillan.

Bremen, H. 1990. No Sustainability Without External Inputs. p. 124-134. In sub-Saharan
Africa beyond adjustment. Africa seminar, Maastricht. Project Group Afiica.
Directorate General for Int. Coop., Ministry of Foreign Affairs, the Hague, the
Netherlands.

Cleaver, K. and G. Schreiber. 1992. The Population, Agriculture and Environment Nexus in
Sub-Saharan Africa. Agriculture and Rural Development Series No. 1, Technical
Department, African Region, World Bank, Washington, DC.

Coulibaly, 0., J. Vitale and J. Sanders. 1998. Expected Efl‘ects of Devaluation on Cereal
Production in the Sudanian Region of Mali. Agricultural Systems, vol. 57, no.4.

12

Dalton, T..1996. Soil Degradation and Technical Change in Southern Mali. Ph.D
Dissertation. Department of Agricultural Economics, Purdue University. West
Lafayette, IN.

Diagana, B., V. Kelly and A. A. Fall. 1995. “Devaluation du Franc CFA et Decisions de
Production Agricole: Une Analyse Empirique de l’Impact de la Devaluation du Franc
CFA sur les Choix de Culture et de Technologies de Production par les Me'nages
Ruraux du Bassin Arachidier du Se'négal” paper presented at PRISAS regional
workshop “Impact de la Devaluation du Franc CFA sur les Revenus et la Securité
Alimentaire en Afiique de l’Ouest”, Bamako, Mali, June.

de Alwis, K. A. 1995. Recapitalization of Soil Productivity in Sub-Saharan AfiicaRome:
FAO Investment Center, Discussion Paper.

Debrah, SK. and W. G. Koster (eds). 1999. Linking Soil Fertility Management to
Agricultural Input and Output Market Development: the key to Sustainable
Agriculture in West Afiica. IFDC-Africa, Lorne, Togo.

Delgado, C., J. Hopkins and V. Kelly. 1994. Agricultural grth linkages in sub-Saharan
Afiica: A synthesis. p. 22-26. In Proc. of a Workshop on Agricultural Growth
Linkages in Sub-Saharan Africa, Washington, DC. 26 May 1994. Int. Food Policy
Res. Inst, Washington, DC.

Donovan, G. and F. Casey. 1998. Soil Fertility Management in sub-Saharan Afiica. World
Bank Technical Paper No. 408

Foth, H. D. 1990. Fundamentals of Soil Sciences. Eighth edition. John Wiley and Sons.

Gebremedhin, B. and G. Schwab. 1998.The Economic Importance of Crop Rotation
Systems: Evidence From the Literature. Michigan State University Department of
Agricultural Economics Staff Paper no. 98-13, August.

Gemer H. and A. Uzo Mokwunye (eds). 1995. Use of Phosphate Rock for Sustainable
Agriculture in West Africa, International Fertilizer Development Center - Afiica,
Lomé, Togo.

Jones, C.A., K.J. Boote, S.S. Jagtap, and J .W. Mishoe. 1991. “Soybean Development” Ch.
5 In J. Hanks and J. T. Ritchie (eds). Modeling Plant and Soil Systems. Number 31
in the series of Agronomy. American Society of Agronomy, Inc. Madison, Wisconsin,
71-90.

Kruseman, G., H. Hengsdijk and R. Ruben. 1993. Disentangling the Concept of
Sustainability: conceptual Definitions, Analytical Framework and Operational

13

Techniques in Sustainable Land Use. AB-DLO/WAU DLV Report No. 2,
Wageningen.

Larson, BA. and GB. Frisvold. 1996. Fertilizers to Support Agricultural Development in
Sub-Saharan Africa: What is needed and Why? Food Policy, vol. 21, no. 6: 509-525.

Matter, A and A Gordon. 1996. Emerging Issues Confronting the Renewable Natural
Resources Sector in SSA. Food Policy 21(2), pp. 229-241, May.

McIntire, J ., and J .M. Powell. 1995. African Semi-arid Tr0pical Agriculture Cannot Grow
Without External Inputs. p. 539-554. In J .M. Powell et al., (eds) Livestock and
sustainable nutrient cycling in mixed farming systems of sub-Saharan Afiica. Vol. H.
Tech. Pap. Int. Livestock Ctr. for Afiica, Addis Ababa, Ethiopia.

McIntire, J. 1986. Constraints to Fertilizer Use in sub-Saharan Afiica. In Management of
Nitrogen and Phosphorus Fertilizers in sub-Saharan A fiica. eds A. U. Mokwunye
and P. L.G. Vlek Proceedings of a symposium. Martinus Nijhoff Publishers.

Mudahar, M. S. 1986. Fertilizer Problems and Policies in sub-Saharan Afiica. InManagement
of Nitrogen and Phosphorus Fertilizers in sub-Saharan A fiica. eds A. U. Mokwunye
and P. L.G. Vlek Proceedings of a symposium. Martinus Nijhoff Publishers.

Oldeman, L.R., R. T. A. Hakkeling and W. G. Sombroek. 1991. WorIdMap of the Status of
Human-Induced Soil Degradation: An Explanatory Note. Wageningen, The
Netherlands and Nairobi, Kenya: International Centre and United Nations
Environment Programme.

Oriade, C.A., and CR. Dillon. 1997. Developments in Biophysical and Bioeconomic
Simulation of Agricultural Systems: A Review. Agricultural Economics. 17: 45-58.

Padwick, G.W. 1983. The Maintenance of Soil Fertility in Tropical Afiica: A Review.
Experimental Agriculture. vol. 19: 293-310.

Reardon, T., C. Barrett, V. Kelly and K. Savadogo. 1999. Policy Reforms and Sustainable
Agricultural Intensification in Africa. Development Policy Review, forthcoming.

Reardon, T., V. Kelly, E. Crawford, B. Diagana, J. Dione, K. Savadogo and D. Boughton.
1997. Promoting Sustainable Intensification and Productivity Growth in Sahel
Agriculture after Macroeconomic Policy Reform. Food Policy, 22(4): 3 17-3 27.

Roberts, W. S. and SM. Swinton. 1996. Economic Methods for Comparing Alternative Crop

Production Systems: A Review of Literature. Reprint from American Journal of
Alternative Agriculture, Vol.11, No. 1.

l4

(I)

Ruben, R., A Kuyvenhoven and G. Kruseman. l998.“Bio-Economic Models for Eco-
Regional Development: Policy Instruments for Sustainable Intensification” 1998
AAEA Preconference workshop, Salt Lake City, Utah.

Sanchez, P.A, K. Shepherd, M. Soule, F. Place, R. Buresh, A-M. Izac, U. Mokwunye, F.
Kwesiga, C. Ndiritu and P. Woomer. 1997. Soil fertility Replenishment in Africa: An
Investment in Natural Resource Capital. In Replenishing Soil Fertility in Africa eds
R Buresh, P. Sanchez and F. Calhoun SSSA Special Publication Number 51.
Madison, “fisconsin, USA: Soil Science Society of America and American Society
of Agronomy.

Sanchez, P.A., A-M. Izac, I. Valencia, and C. Pieri. 1995. Soil Fertility Replenishment in
Afiica: A Concept Note. (Nairobi: ICRAF, and Washington DC: World Bank,
draft).

Scherr, S. J. 1999. “Soil Degradation: A Threat to Developing-Country Food Security by
2020?” IFPRI, Food, Agriculture and The Environment Discussion Paper 27,
Febnrary.

Senegal, Republic of, Ministry of Agriculture.1996. Programme Agricole 1997/98.
December.

Smaling, E., J .J . Stoorvogel and P.Windmeijer. 1993. Calculating Soil Nutrient Balances in
Afiica at Different Scales. 11. District Scale. Fertilizer Research. 35(3): 237-250.

Stocking, M. 1987. Measuring Land Degradation. In Land Degradation and Society, eds. P.
Blackie and H. Brookfield, Methuen, London.

Stoorvogel, J .J., and E.M.A. Smaling. 1990. Assessment of Soil Nutrient Depletion in Sub-
Saharan Afiica: 1983 - 2000. Report 28, Winand Staring Centre, Wageningen.

UNDP/UNECA/FSSDD, 1997. ”Capacity Building in Support of Food Security and
Sustainable Agricultural Development in Sub-Saharan Africa: Towards a
Comprehensive Framework for Future Action." Draft Report, Addis Ababa: May.

Weight, D. and V. Kelly. 1998. Restoring Soil Fertility in Sub-Saharan Afiica: Technical and
Economic Issues. A Policy Synthesis Note for USAID - Africa Bureau, Office of
Sustainable Development. Food Security 11 Project, Dept. of Agricultural Economics,
Michigan State University, October.

Williams, T.O., J .M. Powell and S. Femandez-Rivera. 1995. Manure Availability in Relation

to Sustainable Food Crop Production in Semi-Arid West Afiica: Evidence fiom
Niger. Quarterly-Joumal-of-InternationaI-Agriculture. 34:3 , 248-258.

15

Wong, M.F.T., A. Wild and,A.U. Mokwunye. 1991. Overcoming soil nutrient constraints
to crop production in west Afiica: importance of fertilizers and priorities in SF
research. Fertilizer Research. Dordrecht: Kluwer Academic Publishers. Vol. 29 (1)
p. 45-54, July.

World Bank, IFDC and ICRAF. 1994. Feasibility of Rock Phosphate Use as a Capital
Investment in Sub-Saharan Africa: Issues and Opportunities. (draft)

World Bank and FAQ. 1996. Recapitalization of Soil Productivity in Sub-Saharan Afiica:
Identification and Measurement of Benefits and Costs. Discussion paper.

16

Appendix Al. PRESENTATION OF THE AGRICULTURAL AND
PHOSPHATE PROGRAM IN SENEGAL:

A multi-year “Agricultural Program” has been launched in 1997/98 in Senegal. It
basically consists of the following measures:
+ increase peanut producer price from 132 to 150 CFA/kg;
+ change input price or ease input credit conditions (seeds, fertilizer, etc):
. reduce down payment to 10% of requested loans
. reduce annual interest rate to single digit: 7.5% from 12% the year before;
. eliminate tax on agricultural equipment and
+ distribute P products to farmers under a 4-year publicly-fimded national program.
In Senegal, as part of the Agricultural Program, a phosphate distribution program (PP) has
been launched this last cropping season. Publicly-firnded, it distributes to farmers via CNCR
(council body of federated farmer organizations) local phosphate products (blend of tri
calcium phosphate and phosphogypsum, aluminum phosphate) for basal application to address
P-deficiency in soils. According to .Govemment figures, estimated costs of this program are
around CFA 3.2 billion (almost $5 million)‘. Details of the program implementation are the
following:
+ the Government of Senegal (GOS) orders from local fertilizer manufacturers
(ICS/Senchim, SSPT) phosphate products (by-products and waste);
+ A parastatal (Sonagraines) is asked by GOS to deliver P products from plants to

selected rural community (CR) centers in all regions of Senegal;

 

5 In ‘Le Soleil’, a daily Senegalese newspaper, 19 November 1999.

17

+ CR-level committees (made up by local authorities) reporting to CNCR select
beneficiary villages;
+ Village-level subcommittees select recipient farmers based on following criteria:
. have adequate equipment to incorporate products in soils before rains,
. get seed and NPK fertilizer to use in complement to soil-incorporated P-
product,
. pay 2 CFA/kg , i.e 800CFA for the recommended dose of 400 kg/ha
(product is now given away, free of charge); each recipient farmer is expected
to apply product to 3(?) ha,
. transport product from CR center to village.
+ This 4-year program covers all agroclimatic regions:
.’ year 1: around 50,000 tons for an estimated 117,000 ha;
. year 2-4: 500,000 ha treated annually.
+ Although CNCR is responsible for managing this distribution program (monitoring
tasks are planned), it is not clear whether and how selection criteria will be enforced,
and evaluation tasks will be are performed.
So far, no economic study has been carried out at any level in Senegal to support or justify
this program. Little, if anything, is known about the financial and economic profitability of
this PP which is affected by
+, the time span over which its impacts can be traced out

+ the complementary actions that are proposed to ensure the effectiveness of the PP.

18

ESSAY I:

MICROECONOMICS OF SOIL NUTRIENT REPLENISHMENT: A
THEORETICAL DYNAMIC MODEL OF FARM PRODUCTION BEHAVIOR
1. Problem Statement

Low levels of fertilizer use by farmers on already nutrient-poor and degraded soils in
SSA have been documented in numerous farm production studies. In the face of growing
population pressure on lands, low crop yields, low output production and fast soil nutrient
depletion, the lack of significant fertilizer use seriously puts in jeopardy the option of
intensifying crop production in a sustainable manner to keep up with the region’s increasing
food and fiber needs. Such an intensification is considered inevitable to meet the 4% annual
growth rate in agricultural production necessary to improve conditions of living of millions
of rural populations living in this area (Delgado et al., 1987; Larson and Frisvold, 1996).

Increasing farmers’ use of fertilizer from its current levels is one of the key
components of any overall strategy to achieve a sustainable intensification of agricultural
production in SSA. But, the main problem is how to do it. Answers to this challenging
policy question depend not only on the policy environment, but also on a thOrough
understanding of the microeconomic behavior of farm households. Applying more fertilizer
generates service flows that include increased current and future crop output (to be realized
later because of nutrient replenishment, hence of improved soil fertility), both valued by
individual farmers according to their time preferences, i.e their discount rates. We contend
in this essay that, when making optimal short term fertilizer use decisions, farmers face

intertemporal production tradeoffs coming from the long term soil fertility issue, and how its

19

time-dependent benefits and costs are valued by farmers determine, among other things, these
decisions. Consequently, current as well as firture benefits and costs of this decision must be
accounted for when detemrining optimal fertilizer use decisions and assessing its farm

profitability.

2. Profitability of Fertilizer Use in SSA:
2.1 Mixed evidence:

Conflicting empirical evidence exists on the fertilizer profitability question in SSA,
making it a still hotly debated issue in policy and research circles as well. McIntire (1986)
postulates that low nutrient responses to milletand sorghum, two widely cultivated rainfed
crops, reduce the profitability of fertilizer and thus explain to a large extent its low application
rates in SSA. On the same tone, Sanchez et al. (1997) underscore that fertilizer use on food
crops by smallholder farmers has ofien not been profitable in SSA because of high fertilizer
prices, low producer prices of food crops and risk. Even when profitable, purchase of this
input has been hindered by competing urgent and basic needs for the limited liquidities
available to farmers at the beginning of the cropping season.

In contrast to this assessment, Shapiro et al.(1998) argue that inorganic fertilizer is
the only technically efficient and economically profitable way to overcome soil fertility
constraints in semi-arid West Afiica. In Senegal, research work by Diagana et al. (1995)
based on a static risk-free fann household model for two zones of the Senegalese Peanut
Basin, including various cropping modules and non farm activities showed that the use of

fertilizer-based technologies was not financially profitable at the farm level under prevailing

20

in
is:
usir
lha
mte

12.

use
to C
coni
mos
ofltr
ques

inpl

from
ldOp'

OVer-

behar
lodaji

"OPP:

input and product prices, and thus provided a short run financial justification that corroborates
its observed limited use by farmers in the studied areas. Conversely, Coulibaly et al.,(1998),
using a farm risk-programming model for the Sudanian agroecological zone in Mali, found
that, contrary to conventional wisdom about Afiican farmers diversifying rather intensifying,
intensification of cereal production using fertilizer is financially profitable.

2.2. Dynamic dimensions:

The evidence presented above paints a mixed picture of the profitability of fertilizer
use to SSA farmers, especially those in semi arid areas. Several hypotheses have been offered
to explain this ambiguity, among which the removal of fertilizer subsidies, which has been
common in many SSA countries under the Structural Adjustment Programs of the 19808. But,
most of these profitability studies have been done in a simplified static context, and hence
ofi‘ered only a partial view of the question that did not address two important dynamic
questions for farmers: soil quality and uncertainty, each with important behavioral
implications.

In so far as the concern for the quality of land is reflected in how future crop output
from that piece of land evolves and is subsequently valued by the farmer, it then influences
adoption of fertilizer. This concern is not only formed based on farmer’s observations made
over time of soil quality indicators, mainly the trend of output obtained per unit area, but it
also depends on how much they trade present for future output. As a result, it triggers a
behavioral response by the farmer. The more helshe values tomorrow’s output relative to
today’s, the former depending, among other things, on maintaining the fertility status of the

cropped soils, the more likely he/she is to use fertilizer to restore the soil’s productive

21

capacity, other things being equal. Another view of soil quality separates it from crop
production flows. Van Kooten et al., (1990) included soil quality in the farmer’s utility
firnction to investigate the tradeoffs between net returns and stewardship practices that
require the soil resource to be used so that long term productivity is not impaired, and they
found that a substantial amount of concern for soil quality must be felt before changes in
agronomic practices are observed.

The farming environment in SSA is marked by uncertainty due to erratic rainfall,
unstable prices and unpredictable input distribution policies, etc. Such uncertainty makes
farming activities risky. Farmers deal with this situation by adopting different risk
management strategies that are reflected in their crop, input and technological choices
(Anderson et al., 1977; Sadoulet and DeJanvry, 1995).Young (1979) and Saha (1994) note
a substantial body of evidence from India, Brazil, Mexico, etc that suggests that the typical
farmer in developing countries is risk averse (Moscardi and DeJanvry, 1977; Dillon and
Scandizzo, 1978; Binswanger, 1980). Risk, especially risk aversion, deters adoption of
fertilizer as uncertainty affects the ability of farmers to make good guesses on critical variables
that affect their cropping decisions. Adesina et al. (1988), using a MOTAD risk programming
model, found that the more risk averse farmers in Southern Niger applied fertilizer only on
a limited crop area, and the less risk averse would use more fertilizer, even though cash and
seasonal labor constraints would limit it.

In sum, an evaluation of fertilizer profitability and its use should include these long
term soil quality and uncertainty issues. A good deal of attention has been paid in the adoption

literature to the uncertainty/risk problem which has been dealt with in different empirical ways

22

(“safety-first”, mean-variance, target MOTAD, chance-constrained models, etc.) whereas the
soil quality issue still remains underinvesti gated. Moreover, hardly have the two of them been
explicitly addressed simultaneously. In this essay, we will ignore uncertainty and focus on the
intertemporal tradeoffs involved in the soil quality issue. This analysis is done in a dynamic
fiarnework that reflects the farmer decisionmaking process. Such a process takes place in an
integrated system framework that includes the farm household set of economic activities, its

labor, capital and other resource constraints, its food security needs, etc.

3. Objective of Essay:

The main objective of this theoretical essay is to construct a dynamic multi-sectoral
bioeconomic model of farm households that helps explain how the consideration and the
valuation Of time-dependent service flows (crop production) from soil quality maintenance
or improvement affects optimal crop and technological choices, hence fertilizer demand and
use in an uncertain farming environment. The model is dynamic because it looks at decisions
that have observable effects over multiple periods. It is multi-sectoral as it considers different
activities fi'om different sectors (in and off-farm) that are available to the farm household. It
is bio-economic because it links biophysical crop production processes to economic
management decisions. However, it departs fi'om the standard economic model by explicitly
considering the dynamic production effects of replenishing soil nutrients via fertilizer use as
a way to enhance agronomic sustainability and economic profitability of farming. It does so
by incorporating a soil quality variable as an argument in the crop production firnction and

by keeping track of the motion of the soil asset.

23

Defined as “the inherent capability of the soil to perform a range of productive,
environmental and habitat functions” (Scherr, 1999), the soil quality variable has been
represented in different ways in the farm production literature. Most soil or land quality
studies refer to the productive firnction and are based on an aggregate index of selected soil
physical or chemical characteristics with unknown or subjective weights. For example, to
name a few, Van Kooten et al., (1990) measured soil quality by soil depth and available soil
moisture. Burt (1981) characterized soil quality by topsoil depth and organic matter.
Following those lines, our model uses soil macronutrients (N, P) and organic matter as an

indicator of the quality of the soil resource.

4. Dynamic Bioeconomic Model of Farm Household Behavior

Our theoretical model incorporates a soil quality variable in a-farm household model
that uses the standard expected income maximization approach along the lines presented in
Singh et al. (1986) and Sadoulet and DeJanvry (1995). Biophysical processes of plant, soil
and weather interactions are included in the production part of the model. Nonfarm is
included in the rural farm household’s total set of choice activities.

For farm households assumed to be expected income-maximizers, the multi-period
objective firnction is to maximize the discounted expected stream of income or net farm
returns to cropping and noncropping activities. Mathematically specified, it is as follows:

Max 2,(1+6)“ E (Y,) =- )3, (1+6)“ (Y. + Y,,)

=2r(1+6)"[p*F-(Q.)-r*X.+W.."n...] (1)

subject to

24

where

-<_~<rr

KG?

'9.

qt:

Q, = Q (X. n, NB), production function (2)

L = n, + on, , labor constraint (3)
NBM = NB, + nb, (X,, q,| NB,) , with NB0 specified (4)
expectations operator;

net farm income;

discount rate;

time period;

net cropping income;

net non farm income’;

vector of agricultural output per cropping practice;
vector of variable input (labor excluded) used per cropping practice;
agricultural output price vector;

agricultural input price vector;

total family labor;

labor days devoted to cropping activities and

labor days devoted to noncropping activities;

net returns to a day of noncropping activity;

stock of soil nutrient reserves at beginning of time t;
flow of soil nutrient at time t;

yield of output per cropping practice at time t.

 

1 Nonfarm and noncropping income are used interchangeably throughout the text.

25

Equation (2) describes the crop production process in which yield for a crop is the
result of the interaction ofbiophysical processes with management decisions: amount and type
of inputs used, and soil nutrient balance (NB) and other factors (rainfall, etc.). Equation (3)
assumes away leisure time and states that the fixed family labor stock is allocated between
farming and non farm activities. It follows that

n... = L - n. ' (5)

Equation (5) illustrates one of the ways nonfarm activity affects farming. Via the
labor constraint, it can affect farming negatively by competing with it for the household labor.
Anotheriway not modeled here is that nonfarm activity, a main source of cash income for
rural households in different parts of semi arid West Afiica, helps relax the cash constraint,
and finance capital input acquisition (Reardon etal., 1994; Kelly et al., 1993; Savadogo et
al., 1994; Honfoga, 1999).

Also, with predetermined levels of X, and nu, equation (2) represents a crop response
firnction which depicts a direct relationship between output and soil nutrient conditions,
ceteris paribus. It then allows soil quality to influence the effectiveness of inputs and
technologies (Dalton, 1996).

In equation (4), the levels of the stock of each of the soil macro nutrients at the
begirming of next period t+l are in turn affected by their previous levels (NB), but also by
nutrient inflows fiom current input use (X,) and outflows due to crop yield thru nutrient
uptake (q.).. These stocks are also affected by losses from erosion, leaching, volatization, etc.
This difi'erence equation defined for each soil macronutrient illustrates the recursive nature

of soil condition dynamics as changes in soil conditions between two consecutive periods are

26

determined, controlling for other factors, by input use and crop yield levels in the previous
period.

Rewriting the household problem on a per hectare basis (Q, = q, ) afier substituting
equations (5) and (2) into (1), the present value of a stream of normalized expected net farm
returns to be maximized is

max 2. (1+6): Boo = 2. (1+6): [(p* B we n... NB. ) - r*X. + w. (L - n..)] (6)
subject to

NBM = NB, + nb. (X.- Clrl NB).

This is a dynamic problem with NB as a state variable and X, and n, as control or
instrument variables. It is one of determining the optimal values for X, and n1.. which will, via

equations (2)—(5), imply values for q, and NB,.

5. Solving the Model

Using the Bellman equation specification to solve this dynamic resource allocation
problem, we have

max 2. Y. + ( 1+45)‘l 15. V.” (N13...)
with V,,,, the value function at the beginning of t+1 being equal to BTW, (1+5)“ Y,(X'i , n'd).
This can be written out as

max P‘qr(Xr.n...NB.) -I'* Xr+Wn(L-na)

+(1+5)'l E. [Vin (NB. + nb. (XI: (1.! NED] (7)
Assuming the nonnegativity of the decision variables (X,, n“); one can solve (7) via

the first order conditions:

27

X.: i p ( aq. / aX. ) - r + 0+6)“ E[(BV.+1/5NB.+I) * (anb. I «9on = o (8)

“.3 P (aqt/anl.) - Wu +(1+5)"E.[(<9V.+r /3NB.+1)*(3nb/<9ql) * (fir/311.0] = 0 (9)

6. Discussion of Theoretical Results
6.1 Interpreting the first-order conditions:

Equation (8) typically defines a marginal condition that X, must satisfy. Let’s assume
that X, is reduced to a fertilizer capital input, i.e with nonzero residual factor that is applied
annually (Chiao and Gillingham, 1989). Then, the optimal choice condition stated in equation
(8) reflects intertemporal considerations to be accounted for in making the decision to use
fertilizer. The optimality rule requires that the sum of current (marginal value product or the
first term of the left hand side) and discounted future benefits of applying fertilizer (last term)
be equal to its price, r. The firture benefits are realized because of the replenishment of soil
nutrients. As we know, applying fertilizer under normal conditions increases crop yield output
and helps replenish soil by compensating for extracted nutrients; thus its overall benefits,
ceteris paribus, are not only current yield increase, but also future yields, income, etc. Simply
said, economically rational farmers will engage in replenishing soil nutrient up to the point
where the marginal costs of replenishing nutrients by applying fertilizer are covered by the
current and future marginal benefits. This result implies no value to the farmer to replenish
the soil, except the benefits of increased fixture output.

Equation (8) also illustrates the impacts of having negative soil nutrient balance (or
soil mining): in this case, the third term would be negative. X would then not include

fertilizer, but practices that negatively affect nutrient flows. Then, for that equation to hold,

28

current benefit i.e the marginal value product must be higher. In other words, mining the soil
of its nutrients today to increase output is done at the expense of tomorrow’s output. It
follows fiom this optimal condition, farmers adopting nutrient-mining cropping practices incur
a cost in terms of future output reduction, unless they replenish the soil.

Thus, farmers’ behavior with respect to soil mining or replenishment will depend,
among other things, on how much they trade off the present for the future, i.e, on their time
preferences given by the discount rate. The less myopic preferences they have, the less they
discount the firture, the higher the cost, hence, the more nutrient replacement they will invest
in.

Moreover, high subjective discount rates in excess of social rates of time preferences
can cause farmers to undervalue the costs of soil mining, or equivalently the returns to soil-
replenishing investments (McConnell, 1983). Another reason for this underinvestment by
farmers, beyond the scope of this model, is related to the social emphasis argument of the
imperfect market for environmental goods because of extemalities (social benefits) and/or
imperfect knowledge that leads individual producers to undervalue them and, hence,
underinvest in them.

Equation (9) governs labor allocation within the farm household between on and off-
farrn activities. It states that labor is optimally allocated at the point where the marginal value
product of farm labor equates the returns to nonfarm labor, w“. The last term of this equation
shows the effect of labor allocation on future income via soil conditions: as more labor is

withdrawn for farming to be engaged in nonfarm, the second term is reduced (less crop

29

output, less future benefits), the current productivity of farm labor has to increase for that
equation to hold.
6.2 Predicting the impacts of selected policy measures:

This theoretical model not only captures the biophysical interactions between crop
yields and soil nutrient stocks (equations 2 and 4), but also can be used to describe the
behavior of a rational farmer seeking to maximize returns to farming and having to choose
among difi‘erent cropping practices and mindfirl of the soil fertility impacts‘ of each of them.
His decision rules defined by equations (8) and (9) can be used to trace out the expected
farm-level impacts of selected policy measures. Using the above model, the following effects
can be anticipated.

Changes in input and output prices (increase in p and decrease in r) affect the choice
of input sets (X9 through equation (2). Physical capital transfers such as the phosphate product
give-away program launched in Senegal affect the initial stock of soil nutrients, NBo. Under
this program, phosphate products are distributed to farmers to apply to their cropped soils in
order to remedy their deficiency in the phosphorus nutrient. These combined changes have an
impact on land productivity. Anticipated direct effects on crop production are positive
according to equation (2). Impacts on soil nutrients are traced through equation (4): (i)
positive since changes in X, and NBo affect directly NBM and (ii) negative as increased crop
yields (q,) mean higher nutrient uptake by crop plants from the soil. Therefore, the net effects

i.e the soil nutrient balance will be determined by the magnitude of these opposing effects.

 

2 This may be a strong assumption, but one can argue that while most farmers cannot
measure these impacts, they recognize that certain practices contribute in the long run to
the decline or increase of crop production due to land quality changes.

30

7. Conclusions

This theoretical model conceptualizes farm household decisions in a dynamic multi-
sectoral fi'amework and provides an opportunity to explain farmers’ behavior and to anticipate
possible impacts of policy interventions. By explicitly incorporating in the production process
the dynamic effects of soil quality represented here by soil nutrient replenishment at the farm
level, it explains how optimal i.e expected income-maximizing fertilizer use decisions by
farmers involve considering the incremental production flows in different time periods (as a
result from its application) and also the intertemporal tradeoff between current versus future
output. The magnitude of the tradeoff depends on the discount rate. Farmers with high
discount rates are more likely to mine the soil of its nutrients to get increased output in the
short term at the expense of tomorrow’s output. In contrast, those with lower discount rates

are more likely to adopt soil-replenishing practices that maintain or increase future output.

31

References

Adesina, AA, P.C. Abbott and J .H. Sanders. 1988. Ex-ante Risk Programming Appraisal of
New Agricultural Technology: Experiment Station Fertilizer Recommendations in
Southern Niger. Agricultural Systems, 27:23-35. ‘

Anderson, J. R., J. L. Dillon and J. B. Hardaker. 1977 Agg'gjltugl Decision Anflysis. Arnes,
Iowa: The Iowa State University Press.

Binswanger, H. 1990. Attitudes Toward Risk: Experimental Measurement in Rural India.
American Journal of Agricultural Economics 62:395-407.

Burt, O. R. 1981. “Farm level Economics of Soil Conservation in the Palouse Area of the
Northwest”. American Journal of Agricultural Economics, 63: 83-91.

Coulibaly, O., J. Vitale and J. Sanders. 1998. Expected Effects of Devaluation on Cereal
' Production in the Sudanian Region of Mali. Agricultural Systems, 57, no.4.

Chiao, Y. and A Gillingham. 1989. The Value of Stabilizing Fertilizer under Carry-Over
Conditions. American Journal of Agricultural Economics. May,352-3 62.

Dalton, Timothy. 1996. Soil Degradation and Technical Change in Southern Mali. Ph.D
Dissertation. Department of Agricultural Economics, Purdue University, West
Lafayette, IN.

Delgado, C., J. Hopkins and V. Kelly. 1994. Agricultural growth linkages in sub-Saharan
Afiica: A synthesis. p. 22-26. In Proc. of a Workshop on Agricultural Growth
Linkages in Sub-Saharan Africa, Washington, DC. 26 May 1994. Int. Food Policy Res.
Inst, Washington, DC.

Diagana, Bocar, V. Kelly and A. A. Fall. 1995. “ Devaluation du Franc CFA et Decisions de
Production Agricole: Une Analyse Empirique de l’Impact de la Devaluation du Franc
CFA sur les Choix de Culture et de Technologies de Production par les Ménages
Ruraux du Bassin Arachidier du Sénégal” paper presented at PRISAS regional
workshop “Impact de la Devaluation du Franc CFA sur les Revenus et la Securite'
Alimentaire en Afiique de l’Ouest”, Bamako, Mali, June.

Dillon, J and P. Scandizzo. 1978. Risk Attitudes of Subsistence Farms in Northeast Brazil: A
Sampling Approach. American Journal of Agricultural Economics. 60:425-43 5.

Honfoga, B.G. 1999. Farmers Perceptions of Soil Fertility Resulting fi'om Adoption of
Improved Farm Technologies in Southern Togo. in Debrah, SK. and W. G. Koster

32

(eds). Linking Soil Fertility Management to Agricultural Input and Output Market
Development: the key to Sustainable Agriculture in West Afiica. IFDC-Afiica, Lorne,
Togo.

Kelly, V., T. Reardon, B. Diagana, A. Fall and L. Mcneilly. 1993. ”Final Report for the
IFPRI/ISRA Study of Consumption and Supply Impacts of Agricultural Price Policies
in the Peanut Basin and Senegal Oriental" (2 volumes), submitted to USAID/ Senegal,
September.

Larson, BA. and GB. Frisvold. 1996. Fertilizers to Support Agricultural Development in
Sub-Saharan Afiica: What is needed and Why? Food Policy, vol. 21, no. 6: 509-525.

McConnell, K. 1983. “An Economic Model of soil conservation.” American Journal of
Agricultural Economics. 65: 83- 89.

McIntire, J. 1986. Constraints to Fertilizer Use in sub-Saharan Africa. In Management of
Nitrogen andPhosphorus Fertilizers in sub-Saharan Africa. eds A. U. Mokwunye and
P. L.G. Vlek Proceedings of a symposium. Martinus Nijhoff Publishers.

Moscardi, E. and A. de Janvry. 1977. AttitUdes Toward Risk among Peasants: An
Econometric Approach . American Journal of A gricultural Economics. 59: 7 10-7 16.

Reardon, T., E. Crawford and V. Kelly. 1994. Links Between Nonfarm Income and Farm
Investment in Afiican Households: Adding the Capital Market Perspective. American
Journal of Agricultural Economics. 76(5):264-296.

Sadoulet, E. and A. de Janvry. 1995. Quantitative Development Polig Analysis. The John
Hopkins University Press. Baltimore and London.

Saha, Atanu. 1994. A Two-Season Agricultural Household Model of Output and Price
Uncertainty. Journal of Development Economics. 45: 245-269.

Sanchez, P.A, K. Shepherd, M. Soule, F. Place, R. Buresh, A-M. Izac, U. Mokwunye, F.
Kwesiga, C. Ndiritu and P. Woomer. 1997. Soil fertility Replenishment in Afiica: An
Investment in Natural Resource Capital. In Replenishing Soil Fertility in Africa eds
R. Buresh, P. Sanchez and F. Calhoun SSSA Special Publication Number 51.
Madison, Wisconsin, USA: Soil Science Society of America and American Society of
Agronomy.

Savadogo, K., T. Reardon, and K. Pietola. 1994. Farm Productivity in Burkina Faso:Effects

of Animal Traction and Nonfarm Income. American Journal of Agricultural
Economics. 76: 608-612.

33

Scherr, S. J. 1999. “Soil Degradation: A Threat to Developing-Country Food Security by
2020?” IFPRI, Food, Agriculture and The Environment Discussion Paper 27,

February.

Shapiro, B. and J. Sanders. 1998. Fertilizer Use in Semiarid West Africa: Profitability and
Supporting Policy. Agricultural Systems, 56, no.4.

Singh, 1., L. Squire and J. Strauss (eds). 1986. Agriculggal Household Models. Baltimore:
John Hopkins University Press.

Young, D.L. 1979. Risk Preferences of Agricultural Producers: Their Use in Extension and
Research. American Journal of A gricultural Economics. 6 1 (Dec): 1063- l 070.

Van Kooten, G.C., W. P. Weisensel and D. Chinthammit. 1990. “Valuing trade-ofl‘s between

net returns and stewardship practices: The case of soil conservation in Saskatchewan.”
American Journal of A gricultural Economics. 72: 676-680.

34

ESSAY II:

LAND AND QUALITY PRODUCTIVITY OF SELECTED CROPPING
PRACTICES: AN APPLICATION OF BIOPHYSICAL MODELING TO THE
SAHELIAN CONTEXT OF SENEGAL

1. Introduction

Agriculture production in SSA: especially in its ecologically fragile regions, has been
plagued by low productivity and declining soil fertility levels. These two factors are
interrelated. According to a CIMMYT report (1990),

Soil fertility is determined by a combination of several factors including soil depth,

texture, organic matter content, and nutrient replenishment (Speirs and Olsen, 1992).

The most important components of soil fertility are the nutrients nitrogen, phosphorus,

and potassium. Crop yields cannot increase without necessary nutrient levels, nor can

yields be sustained over time or respond to other inputs such as new seeds and
management practices without adequate levelsof soil fertility.

Thus, a key farm-level issue determinant in stopping soil fertility decline and making
cropping systems sustainable is how to economically replenish soil nutrients mined by crops
(Coulibaly et al., 1998). Three questions emerge from this key issue. The first one is the
extent of soil nutrient mining or depletion without replacement. A second one deals with the
alternative techniques or cropping practices available and/or used by farmers to replenish lost
soil nutrients. The third one concerns the profitability of these practices under the uncertain
physical environment of crop production in SSA.

Soil mining or nutrient depletion has been widely reported as a source of reduced crop

yields in SSA agriculture (Stocking, 1987, Bishop and Allen, 1989; Cleaver and Schreiber,

1992; Speirs and Olsen, 1992), and also as a result of certain cropping practices or land uses.

35

In the empirical farm production literature in SSA, attempts have been made to measure the
extent of soil nutrient mining, using conventional research methods of soil surveys and sample
analyses (Stoorvogel et al., 1993; Van der Pol et al., 1993). But, a more comprehensive
assessment of that problem should account for the close interactions between crop plant,
weather, soil and management decisions through biological processes that govern plant growth
and development (Hanks and Ritchie, 1991; Tsuji et al., 1998). Keeping track of these
dynamic processes under variable rainfall, soil condition, and management conditions requires
tools with broader scope than conventional research methods. Fortunately, simulation models
such as crop grth models are designed to handle such interactions, and provide an
appropriate research tool for estimating crop yield and soil nutrient impacts of alternative
cropping practices.

However, use of crop simulation models in SSA is still rare in the regional literature.
Few applications are the calibration of QUEFTS (Quantitative Evaluation of the Fertility of
Tropical Soils) using data from maize fertilizer trials in Kenya (Smaling and Janssen, 1993),
the use of EPIC (Erosion Productivity Impact Calculator) to assess the impacts of
technological change in southern Mali (Dalton, 1996), and the use of CMKEN, a locally
adapted version of CERES-Maize and a subset of DSSAT in Kenya to simulate crop yields
subject to variable plant populations, cultivars, sowing dates and nitrogen fertilizer rates
(Wafula, 1995). Moreover, the performance of most of these models under the extreme
climatic conditions (total rainfall amount and its spatial and temporal distribution) of the

Western Sahel has seldom been validated to date.

36

The range of technical options available to farmers to improve land quality and
productivity is limited. Crop rotation helps smooth macronutrient consumption across crops,
and leads to higher crop yields than monocropping (Bationo and Lompo, 1999), in addition
to reduced soil erosion, less negative environmental extemalities, better soil fertility and less
need for commercial fertilizer (Gebremedhin and Schwab, 1998). However, with the
breakdown of fallow systems under demographic pressure (Speirs and Olsen, 1992; Gaye et
al., 1996; Dalton, 1996), and the incapacity of manure alone to provide the key for attaining
sustainable yield levels (Williams et al., 1995), fertilization remains the only option. It has
even been argued that substantial grth in external inputs use in the form of inorganic
fertilizers is needed to sustain crop production and agricultural growth in SSA (Breman, 1990;
McIntire and Powell, 1995; Larson and Frisvold, 1996), and to prevent mining of soil nutrients
(Seckler et al., 1991).

But, considering historically low levels of inorganic fertilizer use by farmers in these
regions, between 12-15 kg/ha, one should rightly wonder whether inorganic fertilizer use is
seen by farmers as less profitable than alternative crop production practices, hence explaining
its nonuse. Mixed evidence exists on this question, which has experienced renewed interest and
urgency in policy and research circles afier the lackluster results of the 19803 Structural
Adjustment Programs and the challenge to reverse the downward agricultural productivity
trend and the man-caused soil fertility decline of already nutrient-poor soils.

In Senegal, which lies in the arid western Sahelian area of SSA, observed farming
practices include rotating millet and peanut crops grown usually without fertilizer and with

different plant densities. One of these practices that allegedly aggravates the soil mining

37

problem is the practice of higher than recommended levels of seeding density for peanut
cropping with no fertilization at all. Farmers justify it mostly by the desire to dampen yield
reduction due to soil fertility or seed quality decline or lack of fertilizer (Gaye et al., 1996).
But, agronomists argue it reduces crop yields and soil nutrient levels in the long run (Kelly et
al., 1996). Despite works by Kumar and Venkatachari (1971) who found that closer intra-row
spacings, thus higher seeding densities, led to higher peanut yields than farther spacings in
India, the magnitude of the yield and soil fertility effects of these practices in the Sahelian
areas of West Afiica is generally unknown. As Freud et al. (1997) pointed out in their study
of the ‘peanut crisis’ in Senegal, there is a dearth of research on the evolution of soil fertility

in the areas under peanut cropping and its effects on peanut yields.

2. Research Objectives and Questions

Our research objective is to fill the empirical knowledge gap on the three questions
described above. They all require measuring the long term yield and soil nutrient impacts of
different cropping practices. These measures depend upon a well-calibrated and validated crop
simulation model. This study will specifically seek a) to modify and validate a particular crop
growth model to the sahelian context of Senegal, and b) to use the model results to answer
the following research questions:

1. What are the average long term effects of millet-peanut crop rotation under different

cropping practices (fertilizer application rates and seeding densities) on crop yields and

soil nutrients in the Senegalese Peanut Basin?

38

2. Which of these crop rotation practices is financially more profitable in the long run

under the risky conditions of crop production?

The main hypothesis to be tested is whether the practice of using fertilizer on millet-
peanut rotation is financially more profitable in the long run than other cropping practices

under the variable weather and agricultural production conditions existing in western Sahel.

3. Cropping Systems in the Senegalese Peanut Basin

Most of the agricultural output in Senegal comes from the Peanut Basin, which is “a
vast (sahelian) area of rainfed peanut and millet production that represents 33% of Senegal's
land area, 65% of its rural population, 80% of its exportable peanut production, and 70% of
its cereal production” (Kelly et al., 1996). Agricultural production is mainly done by
srnallholding farmers that use traditional cropping practices characterized by continuous
cultivation of soils without fertilization. Yields for millet(a food crop) and peanut (the main
source of cash income) are low and highly variable following the vagaries of rainfall (see figure
1.1) and policy environment. During the last decade, they seem to have turned downward for
peanut and at best stayed stable for millet (see figure 1.2). Most varieties grown in the Peanut
Basin are short cycle. Dry planting before the first useful rain’ is common for millet as it fi'ees
labor for peanut cropping; re-seeding is done when no rains of 7 mm or more fall in the same
week. Seeding for peanut is done the day after the first usefirl rain. Observed seeding densities

for peanut can vary up to twice the recommended 60 kg/ha of grains.

 

1 Defined as the first rain of 20-25 m not followed by a 25-30 day drought period.

39

Rural household incomes are very low, ranging fiom CFA 30 to 60,000 ($50— 100 US)‘
per adult equivalent per year across agroclimatic zones (Kelly et al., 1993). World Bank Living
Standard Measurement Survey results showed that more than half (58%) of rural households
live below the poverty line and most of these poor are located in the Peanut Basin (Republic
of Senegal/MAIGRS, 1997). One consequence of these low incomes and of the limited cash
flows is that they constrain farmers‘ capacity to make capital investments to restore soil fertility
and allow a transition from traditional systems to more productive/intensive systems.

During the last three decades, input distribution and price policies in Senegal have been
characterized by frequent episodic and incoherent changes that added to the instability of the
production environment (Diagana et al., 1996). The 1980 Structural Adjustment Program
contained a progressive elimination of subsidies (especially for inputs). As a result, fertilizer
use followed a downward trend at the national (annual growth rate of -1.8% from 1965 to
1996; more pronounced decline between 1980 and 1986; see figure 1.3) and farm levels (Kelly
et al., 1996).

An outcome of the interaction of factors above has been the declining fertility status
of most of Senegalese soils: 70% of them are rated fiom ‘poor’ to ‘very poor’ in terms of
suitability to agriculture (Mbodj, 1987). These tropical ferruginous soils are poor in physical
and chemical characteristics. Their topsoil and subsoil textures are characterized by sandy or
clay-sandy layers, high kaolinite dominance in the clay fi’actions, low water holding capacity,
low organic matter and low cation exchange capacity (Sheldon, 1987; Mbodj, 1987), all of

which afi'ect crop response.

 

2 $1.0 US = CPA 600

40

Figure 1.1: Total annual rainfall ln Senegal

 

 

 

 

 

 

 

 

 

 

 

 

6000,
A 50W
5,, A
‘2'
= "\‘7vA "'\ A x —Rainfall
O m \ l \l \ I 1
5 V V V V V —Lineartrend
5 2000 -
E
2 1000
0 T r . . . T
In 6 v- v 0 n O N
a :3: 2-2 a E a; .=: 3?; ;=; a
Year
figure 1.2: Average mlllet and peanut ylelds h Senegal
1200-
1000 it x

 

 

 

a... l'
6... ' VAINWIVFWW‘ + M...

-x- Peanut

 

 

 

 

 

 

 

200
o F V V V T i
00 O '- V O I") 03 N ID
Q Q Q O O) Q O) O) Q
'- v- 1- v- v- v- v- v- I- v- v-

Flgure 1.3: Total lnorgenlc tertlllzer (all form ulas)
eonsumptlon ln Senegal (In mettle tons)

100000

80000
70000

50000
40000
30000
20000
1 0000

Fertllzer consumption

 

DNOFflmVSOI-GIDNOFMW
OOOOOQQOQGOO’Q

Ye er

 

[— Netiond fertiizer coneunption —— L'neer trend

 

Around 70-75% of the 2.25 million ha of cropped lands in 1995 were soil nutrient
deficient (Republic of Senegal/ICS/SENCHIM, 1996). Particularly serious is the high
phosphorus (P) deficiency of most Senegalese soils (Republic of Senegal/MA/“Bureau
Pédologie”, 1995; 1997), as levels of total P in soils are in general low. Where total P is
higher, it is not easily available to plants because of the high fixation capacity of soils.
Regressing or disappearing fallow systems in most areas limit the effectiveness of natural soil

nutrient replenishment mechanisms. This places the P-deficiency as the most important

42

biophysical constraint to increased agricultural production that needs to be addressed

(Republic of Senegal/MA, 1996).

4. Materials and Methods
4.1 Materials:

Two zones of the Peanut Basin, center and southeast, have been chosen for this study
because of the importance of rainfed millet and peanut production, their different weather and
soil characteristics, and the availability of data. A 20-year long (1977-96) series of daily
minimum and maximum temperature, sunshine hours and rainfall data for two locations in
Senegal was collected from the ISRA (Senegalese Agricultural Research Institute)
Bioclimatology Center in Bambey, Senegal. The two weather stations, Bambey and Nioro,
respectively lie in the heart of the center and southeast Peanut Basin9 (see tables 1.1a,b for
their weather characteristics). Rainfall is one of the most limiting factors to agricultural
production in the arid areas of the Senegalese Peanut Basin. The rainy season lasts from June
to October when peanut and millet crops are grown; annual total rainfall averaged 497 mm in
Bambey and 674 mm in Nioro over the 1977-96 period and was highly variable.

Soil laboratory analysis results collected from the “Bureau Pédologie” of Senegal
characterize soils from two sites: Colobane in the center and Dioly in the southeast. These

sites are close to the weather stations above. Common soils are tropical ferruginous type: as

 

3 These two stations are the only ones located in the two studied zones of the Peanut
Basin for which a complete series of climatic data are available.

43

Table 1.1a: Weather Characteristics, Center Peanut Basin, Senegal (Bambey station:

latitude: 14.7N, longitude: 46.5:

MONTHLY AVERAGES (1977-96)

Month Solar Radiation Temperature Total Rainfall
MJ/m2 Max.(°C) Min.(°C) Amount (mm) Days
January 17.8 32.4 16.6 1.6 0.3
February 20.4 34.9 17.9 1.2 0.2
March 22.2 35.8 18.8 0.1 0.1
April 24.0 35.8 19.3 0.2 0.1
May 23.1 36.8 20.7 0.3 0.2
June 20.9 36.2 22.9 31.6 2.5
July 20.8 34.4 24.0 87.5 8.1
August 20.2 33.0 23.8 196.1 14.0
September 19.9 33.4 23.6 155.2 11.9
October 20.3 36.3 22.4 21.0 3.6
November 18.8 36.1 19.2 1.4 0.2
16.9 33.3 17.3 0.6 0.4

Table 1.1b: Weather Characteristics, Center Peanut Basin, Senegal (N ioro station:

latitude: 14.1N, longitude: 46.1;

MONTHLY AVERAGES (1977-96)

Month Solar Radiation Temperature Total Rainfall
MJ/m2 Max! °g1 Min! °C) Amount (mm) Dgs
January 17.7 34.0 15.1 0.0 0.0
February 20.4 36.3 16.4 0.0 0.0
March 22.7 37.2 18.8 0.0 0.1
April 23.5 38.0 20.4 0.0 0.0
May ' 22.8 38.0 22.3 3.5 0.8
June 19.6 36.4 23.9 , 62.7 5.2
July 20.5 33.4 24.0 166.7 11.9
August 20.3 32.1 23.5 235.4 15.8
September 19.6 32.5 23.2 153.9 12.7
October 20.0 34.7 22.5 47.2 4.8
November 18.6 36.4 18.2 2.8 0.2
ficembg 16:7 34.2 15.3 114 0.3

Source: Calculated by DSSAT3.5 weather using data fi'om ISRA Bioclimatology Center,

Bambey, Senegal.

44

illustrated by their physical and chemical characteristics (tables 1 .2a,b), their soil fertility status
is generally rated as low.

A major problem for model validation is the dearth of good quality crop yield data
series. The only source we found available was the “Amelioration Fonciere” research trial
yield data sets” on the same two zones of the Peanut Basin. Since they concern peanut millet
rotation under different rates of NPK fertilization for different years between 1973 and 1982,
they will be used to validate the simulation model and their limits discussed.

4.2 Simulation Experiments:

. For validation purposes, three N and P fertilizer treatments similar to those used in the
collected experimental data are considered for each crop of the millet-peanut rotation: 0-0, 61-
14 and 84-14 kg/ha ofN and P for millet and 0-0, 12-12 and 16-16 kg/ha ofN and P for
' peanut.

Three NPK fertilizer treatments were considered. The control treatment receives no
fertilizer at all and reflects traditional cropping practices. The semi-intensive treatment involves
applying 75 kg of 6-20-10 NPK on‘peanut and 150 kg of 14-7-7 and 100 kg of urea on millet.
The intensive treatment consists of 150 kg of 6-20-10 on peanut and 200 kg of 14-7-7 and of
urea on millet. Other specifications such as seeding and spacing are described in tables A. 1 .a
and A 1.b (in appendix) which summarize agronomic information on these two crops in
Senegal. Based on combinations of these three treatments, six scenarios to be simulated in

each of the two zones of the Peanut Basin are:

 

4 See Kelly, 1988 for a detailed description of these data sets.

45

.8? 205 2: E9: 32: com: 25 can 0% cease—co 05 .8 mummi— 303 .85 ”moan 52. com .8352
8a 8% 2883 can can >20 3.: 202 .maouaomvfi Amwm: so 8 @33— wsma 33.8.8 225 mosetg 88:3 .395 =8 ammo—
.25m Set m_ £23 5.5 “mo—nuts, umu in .Z .8856 an $26 .0 uanco Bu arr—mm— ..awocom ammo—ovum 383m ”ooh—om

 

 

. . l . . . . . . . u .
me.” —.e I «d “an 36 Ah.— _.mm 33 Qw— ca
26 ad I ud— nd— and 3..— N.mm Ema n6— 8
find fin 08d 9: 5.2 36 mm; Nam 5mm n6— we
«N6 in 636 9a 9: and an; :H 5mm «.2 cm
and. a.“ encd 9.: v.2 end 34 Qua Anna n6— 2
nad v6 cucd Non a6 and no; odm h.m~ n6— n

 

    
 
 
  

 

G e 6

.5 Eu 8.228 8% 8589 .253
umo In 5322 .828 bacon 53.—
2.8.80 £590 ism :55 38>»

.05 Noam _a . 23m item 2.53m 5.8m anatotfiagu =em £~.— 03.5.

. . .1 . . . . . . . H .
3”.— od I ad Own o—d hm; _.mm ch Qw— co
mn.~ we I ”A: nd— 26 on; ~.mm Emu «.2 co
mm; Ev :cd 0.: >6— 36 on; ~.mm 5mm n6— we
nag he :96 ed 9: 26 8; :H 5mm 02 cm
mad he 299 Q: v.: :6 3A cdm FAN n6— 2
an; 66. 2?: «on ad 26 $4 92 v.2 n6— n

 

.380 0:330 mam es: .225

 

 

II 1 III I , |||Irl 1'11 I] 1|lllllrl'l l

05920.0, ”3mm .3 cam .38.— Shula.— ..oEoU 823—89820 =em 3.".— 93.18

46

- no fertilizer on either crop (MP-00),

- no fertilizer on either crop, and high seeding density on peanut (MP-HSDO),

- no fertilizer on millet, but intensive scheme on peanut (MP-02),

- semi-intensive on both crops (MP-11),

- intensive on both crops (MP-22), and

- intensive on both crops, but with an initial one-time basal application of a 50-50

blend of 400 kg/ha of tri calcium phosphate and phosphogypsum (MP-2PP).

The first three scenarios are observed practices while the last three are recommended,
but seldom observed in the Peanut Basin. Soil conditions are initialized as of 1991 for the
center and 1993 for the south of the Peanut Basin. Replications are each of the 20-year actual
historical weather sequence (1977-96) that displays a wide range of good and also very severe
climatic conditions, allowing to capture variability in rainfall, temperature and sunlight that
influences year-to-year crop yield outcomes.

4.3 Simulation Methods:
4.3.1 Description of the crop growth simulation model:

The Decision Support System for Agrotechnology Transfer (DS SAT, version 3 .5 with
P) model, a product of the International Benchmark Sites Network for Agrotechnology
Transfer (IBSNAT) is used. DSSAT3.5P contains a set of crop growth simulation models.
The CERES models simulate cereal crops (Jones and Kiniry, 1986) whereas the CROPGRO
models handle legume crops (Wilkerson et al., 1983). Parsch et a1. (1991) describe the
CERES models as a family of physiologically-based crop growth models that simulate the

effects of management and environmental variables on daily dry-matter growth, vegetative

47

and reproductive development, and crop yield. Type of cultivars, planting time and density,
fertilizer levels are among the user-specified management variables. Environmental variables
also specified by the user range from daily weather data to parameters showing the basic
characteristics of the soil profile. Both CERES-Millet and CROPGRO-Peanut models of
DSSAT3.5P will be used in this research.

DSSAT3.5P has been credited with two advantages over other crop growth
simulation models. First, it provides a more detailed accounting of phenological development
and stresses encountered in each phenological stage than other models (Kiniry, 1991; Jones
et al., 1991), hence enabling a more accurate prediction of variation in crop yields from year
to year under different planting dates. Second, it requires only moderate amounts of input
data (Krause, 1992; Chu, 1997). Moreover, initially built with a nitrogen (N) focus, the
addition of a P-component has made it suitable for our purpose of simulating both N and P
dynamics in a country where P is deficient in most soils while significant phosphate deposit
reserves exist.

4.3.2 Model modification and testing:

Three main modifications were made to DSSAT3 . 5P. One concerned some of the soil
parameters. Soil water parameter estimates were not available from the aforementioned
collected soil data sets. Thus, drained upper limit and lower limit were determined on the
basis of textural characteristics of soils in the two zones, following indications in Ratliff et al. ,
(1983). Adjustments were then made to the chemical parameters that control the pace at
which P from the applied NPK fertilizer flows in to the labile P pool, and also the flows

between the labile P and the active and stable P pools. In the experiment files, the harvest

48

mode was changed to vary the level of crop residues carry-over between periods. It was set
to 100% to make all grain and top plant residues harvested, which is consistent with the
observed use of crop residues by farmers for purposes (animal feed, fuel, construction, etc.)
other than their reincorporation in the soil.

Test of millet model: To calibrate the CERES-Millet model, we started with its
ICRISAT, Niger local adaptation to similar sahelian conditions. It was modified and then
tested under different NPK fertilizer levels described earlier, using the 1977 weather year and
soil conditions in the two zones of the Peanut Basin.

One specific modification was to revise the genetic coefficients of cultivars used in the
Niger-adaptation of the model in order to reflect the characteristics of those grown in Senegal
(IBV 8001 and 8004). After many model runs, we adjusted the growth genetic characteristics,
using observed cycle length (physiological maturity date minus planting date) and yield as
calibration tools. " The thermal time from seedling emergence to the end of the juvenile phase
during which the plant is not responsive to changes in photoperiod, P1, was set equal to 75
degree days, the thermal time from beginning of grain filling to physiological maturity or P5
to 400 degree days and PHINT, the Phylochron interval or the interval in thermal time
between successive leaf tip appearances to 60 degree days. All other characteristics were kept
similar to those of the IBOOO44-CIVT cultivar used in Niger.

Test of peanut model: The same procedure was also applied to the CROPGRO-

Peanut model. Unfortunately, no Sahel adaptation of the model was found as the CROPGRO

 

5 After a sensitivity analysis to get harvest maturity date and yield estimates reasonably
close to observed in 1977, changes were finally made first on P1 and P5, and then on
PI-IINT.

49

model was developed for conditions very different from those in Senegal in terms of cultivar,
weather and soil type. Modification was made on the cultivars by adjusting some of the
characteristics of the Spanish and Virginia cultivars that appear to be the closest to the ones

grown in Senegal.‘2

4.4 Data Analysis Methods:

First, for validation purpose, the sequential mode of DSSAT3.5P designed for crop
rotations is used to carry out runs of CERES-Millet and CROPGRO-Peanut under the same
fertilizer treatments as are in the observed experimental crop yield data sets described earlier,
using historical weather data from 1977 to 1982 in each of the two zones in the Peanut Basin.
A graphical analysis of these results is done to demonstrate the model performance.

Second, the validated model is run again for 20 years, but under the six scenarios
described earlier. Under each scenario, the model simulates a 20-year long series of crop
yields and soil nutrient balances. Descriptive statistics of these modeling outputs are
calculated to answer the first research question.

Third, these simulated yields data are used along with historical input and output
prices to construct cumulative probability distribution curves of gross margins over variable

costs per ha for each of the six scenarios.l3 Gross margins are defined here as total value of

 

6 Based on personal communication by O. Ndoye (peanut plant breeder, ISRA/Senegal,
doctoral candidate, Texas, A&M, USA), Spanish could be used to approximate either 55-
437 or Fleur 11 or 73-30; by the same token, Virginia can mirror either 28-206 or GH119-
20.

7 Using either the gross margins or the net returns criterion has been found to lead to
consistent results in terms of ranking different cropping practices (Gebremedhin and

50

yield output minus seed and chemical input (fertilizer, urea, fungicides) costs. In the
distribution of these gross margins are embedded production risks due to weather, hence yield
variability. These probability distributions of the different scenarios or cropping practices are
then ranked using stochastic dominance (SD) ordering method" to determine risk-retum

tradeoffs and answer the second research question.

5. Results, and Discussion
5.1. Model Validation Results:

Overall, the model does a reasonable job of simulating crop yields for millet and
peanut in the two zones of the Senegalese Peanut Basin over the 1977-82 period. Plotted
observed versus simulated yield points are scattered fiom the bottom left to the upper right
corner in figure 1.4, suggesting a 1:1 relationship between predicted and measured
observations. Regressing predicted on observed yield values gave a statistically significant
coefficient of predicted yield equal to .97. A t-test of this coefficient being equal to 1 could
not be rejected at 95%.

The curves depicting the cumulative sum of mean yields of the millet-peanut rotation

under each of the three fertilization levels over the 1977-82 period are shown in figure 1.5.

 

Schwab, 1998; Paudel et al., 1998).

8 As a risk efficiency criterion, SD incorporates information on risk and expected returns
in identifying a management strategy (cropping practice) that maximizes expected utility
(here gross margins) and is preferred by decisionmakers with a known risk attitude (King
and Robison, 1981; Boisvert and McCarl, 1990; Parsch et al., 1991).

51

Prodlctad

30001

 

Figure 1.4: Observed vs simulated yields
(kglha) of a hectare of millet-peanut rotation in
the Senegalese Peanut Basin: 1977-82

 

 

 

 

Observed = 0.97'Predieted
. W 8 0.53
O
0 . . 1
0 1000 2000 3000
Observed

 

 

Yield (kglha)

Figure 1.5: Cumulative sum oi mean millet-peanut yields in
the observed and simulated experiments at three levels at
NPK fertilization in the Center Peanut Basin, Senegal

 

 

 

 

 

10000 -
8000 /
—a—Obs-0
6000 +SIm-0
+Obs-1
4000 —e—SIm-1
—-+—Obs-2
2000 _ +SIm-2

 

 

 

 

 

 

 

52

 

In general, the model estimated the yields accurately at all fertilization levels. A very good job
was done for the control plot yield, with estimates (Sim-0) consistently close to measured
(Obs-0) yields. The medium treatment predictions (Sim-1) show some overestimation in mid
years, but converge to the observed (Obs- 1) at the end. The high fertilization level (Obs-2 vs
Sim-2) exhibits some overestimation in the last years. In light of this, the following points
ought to be mentioned about both the measured and predicted data used for validation:

(i). Few extremes, i.e predictions for millet well below observed levels are
observed in 1977. Low rainfall during that year and a period of drought of 2-3 weeks during
critical periods of the millet plant’s growth caused a serious water stress (.3 to .7 on a scale
of 0 to 1, 1 being the maximum) which has significantly depressed biomass production, hence
millet yields for all fertilizer treatments. This highlights the sensitivity of the model to water
stress, a conditioner 0f the effects of nutrients on plant growth.

(ii). There is some uncertainty in the measured yield data illustrated by the absence of
replications in the experiments, and some oddity observed in 1980 (very low yields for all
treatments, especially for higher NPK fertilizer levels), due maybe to some problems (e. g.,
pest attack) that the model does not consider.

(iii). Distance between weather stations and crop experiment sites, though not
substantial, may also have introduced some spatial variation that may explain differences

between predicted and measured yields.

53

5.2. Simulation Results:
5.2.1. Long term biophysical outcomes of cropping practices:
5.2.1.1 Yield effects":

Simulated yields of six cropping practices in both zones of the Senegalese Peanut
Basin are compared in table 1.3. First, predicted yields fi'om these zones seem consistent with
national averages shown in figure 1.2, mostly between 400-800 kg for millet and 600-1000
kg for peanut in the 1977-96 period. As expected, the rainier southern zone has higher yields
than the drier center under all cropping practices but non-fertilized millet.

Long-term average yields increase with fertilization levels for both crops in the two
zones. Under no fertilization, increasing the peanut seed density adds very little to the peanut
yield compared to the control plot in the long run. Allocating all acquired fertilizer to peanut,
the cash crop for which input credits are more readily available than millet, is a common
practice that also benefits the millet crop following the fertilized peanut with a gain of 239-
266 kg over the control treatment. Results also indicate that participating in the phosphate
distribution program under the recommended conditions, i.e using also NPK fertilizer,

generates a combined millet-peanut yield gain of 125-159 kg/ha over the intensive treatment.

 

15 Simulated results on the water balance components are presented in appendices.

54

.EEmEa cozantfifi 22393 05 have: 22605 Became—La co 3:833 .33 25.95 id 8V

£88 38% :9; How: E
Samoa ”ca umEE ”.8 A3 ”cacao;

«E: at -- owe m: -. ,va la: is 3ng1: as
£8 89 :2 «m: 98 m3 #3. 89 fit 59 89 Km .85
«9 0% mom «8 m8 EN own Em 8a EN EN vmm 5%
89 «cm 22 08 4mm. 0mm 82 mam 3: m8 m8 emu came :58

 

mmv mm mom a F F NON mm 8v mm mom mv F mm F mo F EE
new F van F mmm F mmm mvm mun vow F mum F Nmm F n F F F mom was me
0mm mum vow mm F mm F mm F . F FN 8N mm F FNN 3 F om F >o2m

 

 

 

m FoF mmm mvm mvm Nmm Fom omm mm «mm mom came .250
ca .6 _ ca .8 ca . .8 ca .8 ca .8 Amvca 3::
or: £5 a S cease 3 8 35m... co_.m~___:a
355.3 268:: ._E dob .656 co=m~___to_ @2395 cozm~___t8
0232:. no Snatch oz :0 ..ouEtQ 02 0222:. E90; 02

    
 

   

   

 
 
 

a a. 34.5—
2:an :58.— oB-awoaem o... E 89:09:— wamaaeao 39.3.35... $8928.... Beebe 23.» 5.3.98. tea—=55 an; 03:.

55

5.2.1.2. Soil nutrient effects:

The impacts of the cropping practices on selected characteristics of the soils are
shown in tables 1.4a and 1.4b. Traditional non fertilizer-using cropping practices (MP-00 and
MP-HSDO) contribute more to depleting pools of soil nutrients than other practices. For
plant-available labile P and active P pool, the average annual rate of depletion is higher when
no fertilizer is used. In contrast, the stable P pool is being replenished faster under fertilizer-
using cropping practices as the latter give more yields, thus more roots returned to the soil
that increase organic matter decomposition.

Inorganic N is being replenished under all practices, but faster when more N is
externally applied and more is being fixed by the peanut plant. By contrast, the organic N pool
is decreasing under all practices, and at a faster rate when no fertilizer is used, probably
because of higher mineralization to release the necessary N required by the millet crop plant.

Comparing the two zones, one can note that average annual changes in the stock of
P in the three different pools are almost the same while being very different for organic N.
The organic N pool is decreasing almost twice as fast in the rainier south as the drier center
zone. For the inorganic N, the replenishment is faster under almost all practices in the center

than in the south.

56

Table 1.4a: Simulated average long-term effects of millet-peanut cropping practices
on key soil nutrient pools“ in the Center Peanut Basin, Senegal: 1977-96

 

 

(in kg/ha/year)
CENTER
N0 Semi- Intensive N0 fertilizer No Intensive
fertilizer intensive fertilizer on either, fertilizer on fertilizer on
on either fertilizer on both high wd millet; both with
crop on both crops density on intensive phosphate
peanut on peanut program
_ MPOO MPll MPZZ MPHSDO MPOZ MPZPP
P Labile -1.5 0.9 2.4 -1.5 0.9 2.6
Inorganic -2.5 -2 -1.8 -2.5 -2 -1.7
active
Organic -3.3 -3 -3 -3.3 -3. 1 -3
active
Inorganic 1.6 3.6 4.8 1.5 3.7 5.6
stable
Organic 0.9 1.6 2 0.9 1.6 2.2
stable
N Inorganic 3.5 10 14.9 i 3.7 4.4 14.8
N
Organic -46 -35 -31.5 44.6 -35.2 -30.5

5

Notes: * All figures refer to annual changes in kg/ha (level at end-harvest minus level at
beginning-planting) averaged over the simulation period.
. Initial levels in the Center of the Peanut Basin are respectively:
for labile P, 34 kg/ha;
for inorganic P in the active pool, 71 kg/ha;
for organic P in the active pool, 86 kg/ha;
for inorganic P in the stable pool, 278 kg/ha;
for organic P in the stable pool, 204 kg/ha;
for inorganic N (sum of ammonium and nitrate), 33.5kg/ha;
for organic N, 3333 kg/ha.

57

Table 1.4b: Simulated average long-term effects of millet-peanut cropping practices
on key soil nutrient pools“ in the South Peanut Basin, Senegal: 1977-96

(ill IKE/halve")

 

 

 

No Semi- Intensive N0 fertilizer No fertilizer Intensive
fertilizer intensive fertilizer on either, on millet; fertilizer on
on either fertilizer on both high seed intensive on both with

crop on both density on peanut phosphate
peanut program
MPOO MP1 1 MPZZ MPHSDO NIP-02 MP-2PP
P Labile -1.5 0.2 1 -1.7 -0.1 1.7
Inorganic -2.5 -2.1 -1.9 -2.15 -2.1 -1.8
active '
Organic . -4 -3.7 -3.6 -4 -3.7 -3.6
active
Inorganic 1.1 2.9 4 1 3 4.7
stable
Organic 1.4 2.2 2.7 1.45 2.3 2.9
stable
N Inorganic 3.2 5.9 7.6 3.3 2.7 7.6
N
Organic -89.7 -80.6 -76.5 -88.2 -79.7 -75.5

m

Notes: "' All figures refer to annual changes in kg/ha (level at end-harvest minus level at
beginning-planting) averaged over the simulation period.
. Initial levels in the South of the Peanut Basin are respectively:
' for labile P, 26 kg/ha;

for inorganic P in the active pool, 70 kg/ha;
for organic P in the active pool, 104 kg/ha;
for inorganic P in the stable pool, 274 kg/ha;
for organic P in the stable pool, 245 kg/ha;
for inorganic N (sum of ammonium and nitrate), 30.35 kg/ha;
for organic N, 6348 kg/ha.

58

5.2.2. Stochastic dominance analysis of cropping practices:

The cumulative probability density curves of gross margins over variable costs (labor
not included) of the six cropping practices in the two zones of the Senegalese Peanut Basin
are shown in figures 1.6 and 1.7. Using first-degree stochastic dominance (F SD) analysis, the
graphs clearly indicate that non fertilizer-using practices (Cump-O and Cump-hiO) are
dominated by the others. In effect, at any level of gross margins per ha of millet-peanut used
here as a crude measure of financial profitability, the probability of getting that level or less
is always higher when no fertilizer is used at all. For any practice, this probability is measured
as the vertical distance between any point on the X-axis and the corresponding curve. This
means that, for farmers seeking to minimize the downside risk of getting low gross margins,
FSD ordering suggests using fertilizer over not using it. For example, the likelihood of getting
less than CFA 30,000 per ha of millet-peanut is almost certain (close to 100%) in the two
zones when one does not use fertilizer.

Moreover, in both zones, using the intensive scheme on either the peanut crop only
or on both crops (Cump-02, 22, 2PP) is financially more interesting in the long run than
applying the senri-intensive scheme (Cump-l). Among the intensive fertilizer uses, no
dominance is clearly exhibited by any scheme in neither zone . Finally, without fertilization,
practicing high seeding density on peanut is dominated by applying the recommended seeding
rate for gross margins per ha up to CFA 25,000 /ha in the Center and CFA 40,000 /ha in the

South.

59

Cumulative Probablty

Figure 1.0: Cumulative probability distribution curve
of gross margins over variable costs of a ha of millet-
peanut rotation grown under selected management
practices in Center Peanut Basin. Senegd:1977-08

 

 

 

 

 

 

 

 

 

Gross margins (CFNha)

 

+ amp-0 —a— map-1 + Currp-Z
+ (amp-bio + map-02 —i— Cunp-ZPP

 

 

 

Figure 1.7: Cumulative probability distribution curve
of gross margins over variable costs of a ha of millet
peanut rotation grown under selected maragement

.0 r‘
en .. to

.0
5

Cumulative ProbabIIty
.0
a:

P
on

practices in South Peanut Basin, Senegal: 1977-96

 

 

 

 

 

 

 

 

 

 

Gross margins (CFAIha)

90000
1 05000 1
1 15000 l
125000 ‘

 

+ Currp-O —b— Currp-1 —at— Gimp-2
+ Cunp-hio —o— Currp-OZ —i— Currp-ZPP

 

 

 

6O

In sum, given variability in weather conditions and of yield outcomes, using fertilizer
on millet-peanut rotation is a financially more attractive long term cropping alternative than
not using it at all. In financial terms, practicing high seeding density on peanut is not
supported in the long run by the results of this analysis. This lends some support to the
argument that high peanut seeding density under no fertilization as done by farmers, while
financially profitable in the short-run, is financially less interesting than other practices in the

long term (Kelly et al., 1996; 1998).

6. Conclusions

Crop simulation models provide a good research tool to study the land productivity
as well as the environmental consequences of various cropping practices. The system
approach upon which they are based to link biophysical conditions with behavioral crop
management decisions in order to predict outcomes must, however, be shown to be as close
to real situations as possible to gain wider acceptance in applied research. This study has
shown that a modified, calibrated and validated version of the DSSAT model has performed
reasonably well for millet and peanut cropping under the extreme conditions of western Sahel
prevailing in the Senegalese Peanut Basin. However, more should be done to improve model
adaptation to other areas it was not originally designed for. To accomplish this, the serious
data constraint is to be overcome, and conducting research trials in areas such as those of
western Sahel on major crops under variable physical and management conditions is a good

and necessary step in that direction.

61

Modeling results clearly indicate as expected that, in the long run, observed and
widespread practices of not using fertilizer on millet-peanut rotation:

(1) lead to low crop output,

(ii) contribute more to depleting the soil of the necessary macronutrients N and P, and

(iii) are financially less attractive to farmers than fertilizer-based practices in the
Senegalese Peanut Basin.

These analyses done at the plot level highlight some strong incentives to use fertilizer
and lend support to the view that financially and environmentally sustainable land uses in semi
arid areas are possible through increased inorganic fertilizer use. However, another
determinant of the use of this productivity-enhancing and soil nutrient-replenishing input is
the capacity of farmers to get access to fertilizer at low cost, in a timely manner and when
other competing needs for scarce liquidities are urgent. Going beyond this plot-level analysis
by considering the overall environment in which the farmer operates (which is influenced by
policymaking), his/her resource constraints and other objectives will shed more light on the
financial attractiveness of this input. This can be done in a linear programming farm household

model.

62

References

Bishop, J. and J. Allen. 1989.0n-Site Costs of Soil Erosion in Mali. Environment Department
Working Paper No. 21, World Bank, Washington DC, November.

Boisvert, R.N.and B. McCarl, 1990. Agricultural Risk Modeling Using Mathematical
Programming. Southern Cooperative Series, bulletin No. 356, July.

Breman, H. 1990. No Sustainability Without External Inputs. p. 124-134. In sub-Saharan
Afiica beyond adjustment. Africa seminar, Maastricht. Project Group Afiica.
Directorate General for Int. Coop., Ministry of Foreign Affairs, the Hague, the
Netherlands.

Chu, MC. 1997. Designing Production Contracts to Reduce Agricultural Nonpoint Source
Pollution. Ph.D dissertation. Dpt of Ag. Economics, Michigan State University, East
Lansing, MI.

CIMMYT. 1990. CIMMYT World Maize Facts and Trends: Realizing the Potential of Maize
in Sub-Saharan Afiica. CIMMYT, Harare, Zimbabwe.

Cleaver, K. and G. Schreiber. 1992. The Population, Agriculture and Environment Nexus in
Sub-Saharan Afiica. Agriculture and Rural Development Series No. 1, Technical
Department, Afiican Region, World Bank, Washington, DC.

Coulibaly, O., J. Vitale and J. Sanders. 1998. Expected Effects of Devaluation on Cereal
Production in the Sudanian Region of Mali. Agricultural Systems, vol. 57, no.4.

Dalton, T. 1996. Soil Degradation and Technical Change in Southern Mali. Ph.D
Dissertation. Department of Agricultural Economics, Purdue University, West
Lafayette, IN.

Diagana, B., V. Kelly and M. Kébe'. 1996. “L’Offre Agricole Suite :1 la Devaluation:
Pourquoi Une Réponse Si Faible Au Sénégal?” ISRA/PASE-CILSS/PRISAS Note,
October.

Freud, C., E. H. Freud, J. Richard and P. Therein. 1997. "La Crise de l’Arachide Au Senegal:
Un Bilan-Diagnostic" CIRAD report, January.

Gaye, M and V. Kelly. 1996. L’Utilisation des Terres Agricoles dans le Bassin Arachidier
du Senegal: Facteurs Determinants des Combinaisons Culturales et des Systemes de

63

Rotation. Michigan State University Department of Agricultural Economics Staff
Paper no. 96-32, April.

Gaye M., M. Sene and V. Kelly. 1996. Facteurs Determinants des Densites de Semis de
l’Arachide. Michigan State University Department of Agricultural Economics Stafi‘
Paper no. 96-29, April.

Gebremedhin, B. and G. Schwab. 1998. The Economic Importance of Crop Rotation
Systems: Evidence From the Literature. Michigan State University Department of
Agricultural Economics Staff Paper no. 98-13, August.

Hanks, J. and J. T. Ritchie (eds). 1991. Modeling Plant and Soil Systems. Number 31 in the
series of Agronomy. American Society of Agronomy, Inc. Madison, Wisconsin.

Jones, C. A. and JR. Kiniry (eds). 1986. CERES-Maize: A simulation model of Maize
Growth and Development. College Station, TX: Texas A&M University Press.

Jones, CA, K.J. Boote, S.S. Jagtap, and J .W. Mishoe. 1991. “Soybean Development” Ch.
5 In J. Hanks and J. T. Ritchie (eds). Modeling Plant and Soil Systems. Number 31
in the series of Agronomy. American Society of Agronomy, Inc. Madison, Wisconsin.
1991: 71-90. '

Kelly, V., B. Diagana, M. Gaye, T. Reardon and M. Séne. 1998. "L’Ajustement Structure]
Est-i1 Incompatible Avec Une Gestion Durable de la Fertilité des Sols: L’Experience
Senegalaise dans le Bassin Arachidier" in G. Renard, A. Neef, K. Becker and M. Von
Oppem (eds) “Soil Fertility Management in West Afiican Land Systems” Proceedings
of the University of Hohenheim (Germany), ICRISAT Sahelian Center and INRAN
regional workshop, 3-8 March 1997 Niamey, Niger. Werkersheim: Margraf Verlag.
1998.

Kelly, V., B. Diagana, T. Reardon, M. Gaye and E. Crawford. 1996. Cash Crop and
Foodgrain Productivity in Senegal: Historical View, New Survey Evidence, and
Policy Implications, Michigan State University International Development Paper no.
20.

Kelly, V., T. Reardon, B. Diagana, A. Fall and L. Mcneilly. 1993. "Final Report for the
IFPRI/ISRA Study of Consumption and Supply Impacts of Agricultural Price Policies
in the Peanut Basin and Senegal Oriental" (2 volumes), submitted to USAID/ Senegal,
September.

Kelly, V..1988. Factors Affecting the Demand for Fertilizer in Senegal’s Peanut Basin.
Unpublished PhD thesis dissertation, Dpt of Agricultural Economics, Michigan State
University, East Lansing, MI.

King R P. and L. Robison. 1981. Implementation of the Interval Approach to the
Measurement of Decision Maker Preference. Research Report 418, Michigan State
University Agricultural Experiment Station, Michigan State University, November.

Kiniry, JR 1991. “Maize Phasic Development”. Ch. 4 In J. Hanks and J. T. Ritchie (eds).
Modeling Plant and Soil Systems. Number 31 in the series of Agronomy. American
Society of Agronomy, Inc. Madison, Wisconsin. 1991: 55-70.

Krause, M. A 1992. Optimal adoption strategies for conservation tillage technology in
Michigan. Ph.D dissertation. Dpt of Ag. Economics, Michigan State University, East
Lansing, MI.

Kumar, MA. and A. Venkatachari. 1971. Studies on the Effects of Intra Row Spacings and
Fertility Levels on the Yield and Quality of two Varieties of Groundnut (Arachis
Hypogaea L.)”. IndianJ. of Agric. Res. 5(2):67-73.

Larson, BA. and GB. Frisvold. 1996. Fertilizers to Support Agricultural Development in
Sub-Saharan Afiica: What is needed and Why? Food Policy, vol. 21, no. 6: 509-525.

Mbodj, M.1987. L’Utilisation des Engrais et la Production Agricole: Fertilisation des
Cultures au Senegal et dans les Autres Pays de l’Afiique de l’Ouest. Mimeo, Ministry
of Rural Development, Dakar, Senegal.

McIntire, J., and J .M. Powell. 1995. Afiican Serni-arid Tropical Agriculture Cannot Grow
Without External Inputs. p. 539-554. In J .M. Powell et al., (eds) Livestock and
sustainable nutrient cycling in mixed farming systems of sub-Saharan Afiica. Vol. H.
Tech. Pap. Int. Livestock Ctr. for Afiica, Addis Ababa, Ethiopia.

Parsch, L.D., M.J. Cochran, K.L. Trice, and H. D. Scott. 1991. “Biophysical Simulation of
Wheat and Soybean to Assess the Impacts of Timeliness on Double-Cropping
Economics”. Ch. 22 In J.‘ Hanks and J. T. Ritchie (eds). Modeling Plant and Soil
Systems. Number 3 1 in the series of Agronomy. American Society of Agronomy, Inc.
Madison, Wisconsin. 1991: 511-533.

Paudel, K., N. Martin, G. Wehtje, and T. Grey. 1998. “Economic Decision Making Using

Enterprise Budgeting and Statistical Analysis: An Illustration in Weed Control
Practices in Peanut Production.” Journal of Production Agriculture, 11(1):48-73.

65

Ratlifi‘, L.F., J. T. Ritchie and D. K. Cassel.1983. Field-Measured. Limits of Soil Water
Availability as Related to Laboratory-Measured Properties. Soil Science Society of
America Journal, vol. 47, no. 4, July-August.

Seckler D., D. Collin and P. Antoine. 1991. Agricultural Potential of “Mid Afiica”: A
Technological Assessment in Agricultural Technology in Sub-Saharan Afiica, eds S.
Gnaegy and IR. Anderson, pp. 61-103, World Bank Discussion Paper 126, World
Banlg Washington, DC.

Sene, M. 1995. Influence de l’Etat Hydrique et du Comportement du Sol sur l’Implantation
et la Fructification de l’Arachide. These doctorat en Agronomic. Ecole Nationale
Superieure Agronomique, Montpellier, Institut National de la Recherche
Agronomique, Laon, France.

Senegal, Republic of, Ministry of Agriculture. 1996. Programme Agricole 1997/98.
December.

Senegal, Republic of, Ministry of Agriculture, “Groupe de Reflexion Strategique”. 1997.
Rapport sur La Politique Agricole au Senegal.

Senegal, Republic of, Ministry of Agriculture, Bureau Pédologie. 1997. Programme de
Restauration de la Fertilité des Sols du Sénégal. Avril.

Senegal, Republic of, Ministry of Agriculture, Bureau Pédologie. 1995. Rapport Annuel.

Senegal, Republic of, ICS/SENCHIM. 1996. Projet de Restauration de la Productivite des
Sols au Senegal. Mimeo, Octobre.

Senegal, Republic of, Ministry of Energy, Mines and Industry. 1996. Projet de Fertilisation
des Sols et d’Utilisation des Engrais Phosphates. October.

Sheldon, V.1987. Fertilig Status of African Soils: a Handbook for Non Aggonomists.

Curriculum Publications Clearinghouse, Western Illinois University.
Smaling E.M.A. and B.H. Janssen. 1993. Calibration of QUEFTS, A Model Predicting

Nutrient Uptake and Yield from Chemical Soil Fertility Indices. Geoderma. 5921-4,
21-44.

66

Speirs M. and 0. Olsen. 1992. Indigenous Integrated Farming Systems in the Sahel. World
Bank Technical Paper No. 179, Afiica Technical Department Series, World Bank,
Washington, DC.

Stocking, M. 1987. Measuring Land Degradation. In Land Degradation and Society, eds. P.
Blackie'and H. Brookfield, Methuen, London.

Stoorvogel, J .J ., E.M.A. Smaling and B.H. Janssen. 1993. Calculating Soil Nutrient Balances
in Afiica at Different Scales. I. Supra-national Scale. Fertilizer Research. 3 5(3): 227-
235.

Tsuji, G.Y. et al., (eds). 1998. Understanding Options for Agricultural Production. Kluwer
Academic Publishers.

Van Der Pol, F. 1992. Soil Mining: an Unseen Contributor to Farm Income in Southern Mali.
Royal Tropical Institute, Amsterdam.

Wafirla, BM. 1995. Applications of Crop Simulation in Agricultural Extension and Research
in Kenya. Agricultural Systems, 49: 399-412.

Wilkerson, G. G., J.W. Jones, K]. Boote, K.T. Ingram, and J .W. Mishoe. 1983. “Modeling
Soybean Growth for Crop Management.” Trans. ASAE, 26(1983):63-73.

Williams, T.O., J .M. Powell and S. Femandez-Rivera. 1995. Manure Availability in Relation

to Sustainable Food Crop Production in Semi-Arid West Africa: Evidence from
Niger. Quarterly-Joumal-of-InternationaI-Agriculture. 34:3 , 248-258.

67

APPENDIX A2

Table A.2.a: Agronomic characteristics of millet production in the Peanut Basin (PB)
a

llet ltivar una 3 IB 8001 IBV 8004
Cycle length (days) 85 - 95 75 - 95 75 - 95
Seeding density 30 000 30 000 30 000
(plant/ha)

Plant spacing (cm) 100 x 100 100 x 100 100 x 100
Rainfall zones (mm) 2 400 2400 300 - 400
ner uhPB ntr o hPB orthPB

Source: M.Sene, agronomist, ISRA/Bambey, Senegal, personal communication, January
1998.

. roduction in the Peanut Basin ’ B

   

 

 

Table A.2.b: A ' . nomic characteristics of nut

      

Peanut Cultrvar 28-206 & 73-33 Fleur 11 55-437 &

 

ng 19-20 73-30
Cycle length (days) 120 110 90 90
Seeding density 110 000 130 000 135 000 166 000
(plant/ha)
Plant spacing (cm) 60 x 15 50 x 15 50 x 15 40 x 15
Rainfall zones (mm) 2700 400 - 700 300 - 500 300 - 500
Southwest South PB Center PB Center north
PB north PB

Source: M. Sene, agronomist, ISRA/Bambey, Senegal, personal communication, January
1998.

68

 

APPENDIX A3

Table A3.1.a: Summary of simulated water balance component results in
Center Peanut Basin for no-fertilizer treatment

 

Year Total Rain in Total . Total Total Change in stored
rainfall crop runoff drainage Evapotransp. water
mm season % mm mm mm mm
m
77 389 83.3 22 0 297 70
78 794 78.3 105 62 525 102
79 445 90.8 35 47 405 -42
80 404 93.8 32 83 309 -20
81 504 97.6 1 12 34 350 8
82 454 96.5 37 72 357 -12
83 434 58.9 56 0 318 60
84 342 96.5 39 37 323 -57
85 390 94.1 15 21 337 17
86 425 84.9 61 62 290 12
87 378 62.7 13 45 317 3
88 648 94.3 123 204 315 6
89 799 93.1 165 289 350 -5
90 392 88.2 36 18 376 -38
91 371 85.2 37 33 268 33
92 339 98.2 32 67 273 -33
93 500 93.2 85 126 255 ' 34
94 484 90. l 79 94 344 -33
95 587 94.9 66 , 171 319 31
96 310 100 . 21 22 281 -14
Mean 469.4 88.7 58.55 74.35 330.45 6.1

_
Source: DSSAT3.5P outputs.

69

Table A3.1.b: Summary of simulated water balance component results in
Center Peanut Basin for semi-intensive fertilizer treatment.

Year Total Rain in crop Total Total Total Change in
rainfall season runoff drainage Evapotransp. stored water
mrn % mm mm mm mm
77 389 83.3 22 0 309 58
78 832 75.3 1 10 66 545 111
79 407 89.9 32 29 395 -49
80 429 88.3 33 64 334 -2
81 ' 479 97.5 112 26 360 -19
82 529 82.8 60 45 385 39
83 359 50.4 32 0 329 -2
84 373 88.5 39 3 370 -39
85 359 93.6 15 0 358 - -14
86 , 425 96.7 62 22 286 55
87 378 62.7 13 0 364 1
88 710 86 134 180 365 31
89 737 92.5 160 254 365 -42
90 392 88.3 37 0 412 -57
91 371 85.2 37 0 271 63
92 340 97.9 33 31 326 -50
93 499 93.2 85 66 295 53
94 565 77.2 104 81 390 -10
95 . 507 94.1 46 110 341 10
96 305 100 22 8 320 -45
Mean 469.2 86.2 59.4 49.25 356 4.6

_
Source: DSSAT3.5P outputs

70

Table A3.2.a: Summary of simulated water balance component results in
South Peanut Basin for no-fertilizer treatment

Year Total Rain in crop Total Total Total Change in stored
rainfall season runoff drainage Evapotransp. water
mm % mm mm mm mm
—
77 567 60.5 46 80 299 142
78 772 81.9 79 127 563 3
79 709 86.7 82 167 485 -25
80 519 92.9 66 47 462 -56
81 782 83.4 66 218 442 56
82 545 91.4 33 47 500 -35
83 510 69.2 74 10 383 43
84 442 99. 1 52 35 420 -65
85 531 78.9 32 36 406 57
86 ‘ 876 90.2 168 193 489 26
87 833 68.8 91 337 432 -27
88 956 95.1 178 294 452 32
89 795 81.2 79 268 469 -21
90 536 90.5 52 67 480 -63
91 545 67.5 47 104 319 75
92 695 97.4 99 173 466 -43
93 ‘ 789 79.7 1 1 1 246 403 29
94 721 94.7 88 179 468 -14
95 696 81.7 77 246 3 71 2
96 487 100 67 64 405 -49
Mean 665.3 84.5 79.3 146.9 435.7 3.3

E
Source: DSSAT3.5P outputs

71

 

Table A3.2.b: Summary of simulated water balance component results in
South Peanut Basin for semi-intensive fertilizer treatment

Year Total Rain in Total Total Total Change in
rainfall crop season runoff drainage Evapotransp. stored water
mm % mm mm mm m
“=3
77 567 60.5 45 79 300 143
78 772 81.9 82 125 564 1
79 709 86.7 84 152 497 -24
80 5 19 92.9 69 40 472 -62
81 782 83.4 67 196 456 63
82 545 91.4 36 40 511 -42
83 510 69.2 72 0 408 30
84 442 99.1 53 12 433 -56
85 531 78.9 33 0 452 46
86 876 90.2 175 155 499 47
87 833 68.8 92 286 480 -25
88 956 95.1 186 283 456 31
89 795 81.2 80 217 517 -19
90 536 90.5 55 46 506 -71
91 545 67 .5 46 63 353 83
92 695 97.4 106 139 495 -45
93 789 79.7 113 191 453 32
94 721 94.7 94 167 480 -20
95 696 81 .7 80 202 406 8
96 487 100 70 40 440 -63
Mean 665.3 84.5 81.9 121.6 458.9 2.8

—
Source: DSSAT3.5P outputs

72

ESSAY III:

MICROECONOMICS OF SOIL NUTRIENT REPLENISHMENT IN
SUBSAHARAN AFRICA: EVIDENCE FROM SENEGAL AND PROSPECTS
FOR A SUSTAINABLE FARM IN TENSIFICATION

1. Introduction

Under adverse agroecological conditions, especially low and erratic rainfall, nutrient-
poor and degraded soils, water and soil erosion, agricultural production in Sub-Saharan
Afiica (SSA) has grown very sluggishly in the last three decades, well under the rapid pace
of demographic grth at around 3% per year. To meet soaring food and fiber needs fi'om
a fast growing population, it has been argued that agricultural production should grow at an
estimated rate of 4% per annum. This would then require an annual increase of 1.5% fOr
labor productivity and of 3% for land productivity (Delgado et al.,1987; Cleaver and
Schreiber, 1992; Larson and Frisvold, 1996).

The debate about production path options to meet this serious agricultural
productivity grth challenge has raised several issues. First, extensification onto new and
marginal lands offers limited potential to increase production, and is even likely to put further
pressure on forested areas and resources, leading to more land degradation and deforestation
(Marter and Gordon, 1996). Second, intensification paths which involve using more
productivity-enhancing inputs per unit of land area (improved seeds, chemical inputs, labor)
can be of various types differentiated by their productivity and soil fertility impacts (Reardon

et al., 1997;1999). Whatever the production path, there is a growing consensus that the

73

appropriate one capable of meeting the productivity challenge must be also sustainable, i.e
with a real potential to increase crop yields, to ensure a concomitant and appropriate
replenishment of soils and to generate profits to the farmer.

Despite all existing evidence underscoring the fact that inorganic fertilizer is a key
element for sustainable land use and crop production (Mudahar, 1986; Padwick, 1983; Larson
and Frisvold, 1996; Shapiro et al., 1998), resource-poor farmers in SSA use this input very
sparsely, at levels (12-15 kg/ha) well below world standards. Justification for this behavior
can be sought in general through the lack of strong incentives and/or capacity to acquire and
use this input. Both of these two causes can be altered thru policymaking informed by sound
and relevant research. Under market-oriented reforms, changes in the structure of product and
factor prices, credit conditions and capital transfers are made to affect incentives faced by and
the capacity of farmers to invest in this input. An example is the ‘Agricultural Program’
launched in Senegal in 1997/98 which contained among others the following policy measures:
peanut output price increases, reduction of downpayment requirements on input credits,
distribution of phosphate products to farmers to remedy their deficiency in the phosphorus
(P) nutrient.

Also afi‘ecting fertilizer use are two factors of concern to the farmer, soil quality and
uncertainty, with important behavioral implications. In so far as the quality of land is reflected
in how future crop output from that piece of land evolves and is subsequently valued by the
farmer, concerns for it then influences adoption of fertilizer. This concern, based on farrners’

observations made over time of the trend of output obtained per unit area, also depends on

74

how much they trade present for firture output, i.e on their discount rate. The more they value
tomorrow’s output relative to today’s, the former depending, among other things, on
maintaining the fertility status of the cropped soils, the more likely they are to use fertilizer
to restore the soil’s productive capacity, other things being equal.

Uncertainty due to variable weather, unstable prices and unpredictable input
distribution policies makes farming activities risky in SSA. Farmers deal with this situation
by adopting different risk management strategies that are reflected in their crop, input and
technological choices. It is Often suggested that the typical farmer in developing countries is
risk averse (Moscardi and DeJanvry, 1977; Dillon and Scandizzo, 1978; Binswanger, 1980).
Evidence exists that risk aversion deters adoption of fertilizer. Adesina et al. (1988), using
a MOTAD risk programming model, found that the more risk averse farmers in Southern
Nrger applied fertilizer only on a limited crop area, and the less risk averse would use more
fertilizer, even though cash and seasonal labor constraints would limit it.

However, the available evidence on the dynamic effects of policies on the farm
profitability of cropping practices that condition production paths has so far been scarce and,
above all, inconclusive. Moreover, in the empirical literature, fertilizer profitability studies
in SSA have usually considered the positive effect of fertilizer use on current production and
returns to land while paying little attention to the very often important effects on future
production flows due to maintained or improved land quality. The empirical neglect of land
quality outcomes in these analyses has limited the extent of their contributions to the

sustainable agricultural intensification policy debate in SSA Freud et al., (1997) point to the

75

same flaw when they deplore the lack of empirical evidence about the "consequences of the
elimination of the fertilizer distribution program on soil fertility and crop yields in Senegal".
Crucial to determining these dynamic effects is deciphering how they are mediated through
the farmer’s response (in terms of activity and technological choices) to policy-driven market
signals.

The main objective of this article is to bridge the empirical knowledge gap highlighted
above. It first attempts to build a farm household model that includes various cropping
practices and their dynamic effects on soil fertility and then uses the model to shed light on
two important policy questions.

(i)Will recent price, credit and capital transfer policy changes in Senegal encourage

farmers in the Peanut Basin in the long run to intensify crop production by adopting

fertilizer-using cropping practices or not?

(ii) Subsequently from the model solution, what will the corresponding effects of the

A optimal crop production practices be on soil fertility, here proxied by the soil
macronutrient (N and P) stocks?

We contend that, if current policies mentioned above stand in Senegal during the next
decade with the same trend for input and output prices, incremental yields due to increased
fertilizer use would increase profits to farmers, hence offering them incentives to invest in this
input. But, initial capital constraints would have to be overcome to increase its use, and
hence to improve soil fertility conditions. Ways to ease capital constraints and finance

fertilizer acquisition can be increased access to nonfarm activity and less restrictive formal

76

credit requirements (for example reduced downpayment). Increased participation in nonfarm
activity, a source of cash income, can influence long term crop and input choices (especially
fertilizer use) by severely cash-constrained farmers, but it can also limit the available farm
household labor supply to farming.

The paper is organized along the analytical steps described in figure 1 presented earlier
in the general introduction. First, a set of cropping practices is identified in the study zones
and their impacts on crop yields and nutrient stocks are estimated in Senegal with a crop
grth simulation model using weather and soil condition inputs. Then, these simulated
results are plugged in a multi-period farm household model as production coefficients. This
household model also incorporates information on policy variables such as prices and credit
requirements along with the household nonfarm activities and its resource endowment set.
Finally, linear programming (LP) is used to solve the multi-period household problem for the
optimal set of cropping and nonfarm activities that maximizes the farm household’s objective

function.

2. Cropping Practices in the Senegalese Peanut Basin
2.1 Context:

Farmers in the Senegalese Peanut Basin are mainly millet and peanut producers under
rainfed conditions. This sahelian area represents 33% ofSenegal's land area, 65% of its rural
population, 80% of its exportable peanut production, and 70% of its cereal production.

Rainfall is a serious limiting factor to agricultural production. The rainy season lasts from June

77

to October when peanut and millet crops are grown. Annual total rainfall averaged 497 mm
in the center and 674 mm in the south of the Peanut Basin over the 1977-96 period and was
highly variable. Yields for millet (a food crop) and peanut (the main source of cropping
income) are low and fluctuate a lot with the vagaries of rainfall and the instability of the
policy environment. Low rural household incomes between CFA 30 to 60,000 (around $50-
100 US) per adult equivalent per year across agroclimatic zones (Kelly et al., 1993) and
limited cash flows constrain farmers' capacity to make capital investments to restore soil
fertility and allow a transition from traditional systems to more productive/intensive systems.
During the last three decades, instability in input distribution and price policies in
Senegal has compounded the uncertainty of the production environment (Diagana et al.,
1996). Added to the progressive elimination of subsidies (especially for inputs) under the
19805 Structural Adjustment Program, this has led to a decline of fertilizer use at the national
(annual grth rate of -1 .8% from 1965 to 1996) and farm levels (Kelly et al., 1996).
Most Senegalese soils have low soil fertility status: 70% of them are rated from ‘poor’
to ‘very poor’ in terms of suitability to agriculture (Mbodj, 1987). Their topsoil and subsoil
textures are characterized by sandy or clay-sandy layers, high kaolinite dominance in the clay
fi‘actions, low water holding capacity, low organic matter and low cation exchange capacity
(Sheldon, 1987; Mbodj, 1987), all of which affect crop response. In addition, three-quarters
of the 2.25 million ha cropped in 1995 were said to be nutrient deficient (Republic of
Senegal/ICS/SENCHIM, 1996). Particularly serious is the high P-deficiency of these soils

(Republic of Senegal /MA/Bureau Pédologie, 1995; 1997), a problem diagnosed as the most

78

important biophysical constraint to increased agricultural production (Republic of
Senegal/MA, 1996).
2.2 Cropping practices:

Peanut and millet are usually grown in rotation in the Peanut Basin and most varieties
are short cycle. Dry planting before the first usefirl rain is common for millet as it frees labor
for peanut cropping. Seeding for peanut is done the day after the first usefirl rain. Observed
seeding densities for peanut can vary up to twice the recommended 60 kg/ha of grains.
Farmers justify this high seeding density for peanut by the desire to dampen yield reduction
due to soil fertility or seed quality decline or lack of fertilizer (Gaye et al., 1996). But,
agronomists argue that, without fertilizer, high seeding density would depress crop yields and
soil nutrient levels in the long run (Kelly et al., 1996).

Farmers, mostly smallholders, predominantly use traditional cropping practices
characterized by continuous cultivation of soils without fertilization. Research- recommended
fertilizer application rates are 75 kg of 6-20-10 NPK on peanut and 150 kg of 14-7-7 NPK
and 100 kg of urea on millet (semi-intensive scheme) and 150 kg of 6-20-1 0 NPK on peanut
and 200 kg of 14-7-7NPK and of urea on millet (intensive scheme). When farmers do use
fertilizer, it is usually on the peanut cash crop. Moreover, a phosphate program has been
recently launched: phosphate products are being distributed nationally and free of charge to
farmers to apply to their fields in addition to using NPK fertilizer in order to correct the soil

P-deficiency mentioned earlier.

79

q

2.3 Simulated biophysical outcomes:

Using these observed or recommended practices on millet-peanut rotation, we have
simulated with the DSSATl6 model their long-term yield and soil nutrient effects under six
scenarios in two zones of the Peanut Basin:

- no fertilizer on either crop (MP-00),

- no fertilizer on either crop, but high seeding density on peanut (MP-HSDO),

- no fertilizer on millet, but intensive scheme on peanut (MP-02),

- semi intensive on both crops (MP-11),

- intensive on both crops (MP-22), and

- intensive on both crops, but with an initial one-time basal application of a 50-50

blend of 400 kg/ha of tri-calcium phosphate and phosphogypsum (MP-2PP).

Results are summarized in table 3.1 below. Long term average yields increase with
fertilization levels for both crops in the two zones. Under no fertilization, increasing the
peanut seed density adds very little to the peanut yield compared to the no-fertilizer case in
the long run. Allocating all acquired fertilizer to peanut benefits the millet crop following the
fertilized peanut with a gain of 239-266 kg over the no-fertilizer case. Also, participating in

the phosphate distribution program under the recommended conditions, i.e using also NPK

 

3 DSSAT, the Decision Support System for Agrotechnology Transfer, a product of the
International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT) is a set
of crop growth simulation models. Its CERES models simulate cereal crops whereas the
CROPGRO models handle legume crops (Hanks and Ritchie, 1991).

80

fertilizer, generates a combined average millet-peanut yield gain of 125-159 kg/ha which is
probably not enough to cover the full P costs to the farmer.

Another overall result from this table is that, as expected, traditional non fertilizer-
using cropping practices (MP-00 and MP-HSDO) deplete the stocks of soil nutrients more
than the fertilizer-using practices. For the plant-available labile P pool, the stock is being
depleted when no fertilizer is used, and replenished faster with increased levels of fertilizer
application. In contrast, the active pool of P is being depleted by net flows of P into the labile
pool, but at a slower pace when fertilizer is used. Inorganic N is being replenished under all
practices, but faster when more N is externally applied. By contrast, the stock of organic N
is going down under all practices, and more rapidly when no fertilizer is used, probably

because of higher mineralization to release the necessary N required by the millet crop plant.

81

 

.368... «5:2... ..o 5:030: a 633%... :5. 4.3-2:... a 83%.:

2.23.. “8985 .35... omega 0.... 85w: .2 2.8m... :5 oanBE ..ooa 9:2. 2: E m 0.5%... :5 2.830.: .m 25..— S...
3531.... 305..."? ”geese. 553-552 6.5m“..—

mne—

n_o_

 

an

.annrznueamresn
aza.arzuaix_
neua=.:m_o>::aua

 

en.
es.

wdw

mg.
n;
moo
“wan.
ad;
n-

#4.

ON

nnu

a:

on.
sun

FAV

fin.
n—AY
chm—
«fin.
ed.
fin.
N.
66
av»

an

sannn.nean¢
o>aeuuauo=1=
ue.na=:h¢.;a

can

«an

en

Wm

@661

“an

fin.

can

   

enn—

n;n-

o6—

on»

a:

arzungx.ee

.w.

fin

66.

Ne

mm.

:—

owe

:e

axzunix.eooa=

“35.8n::nuo>vnsa_ nagwnuo>flcsnfiwaow

 

man

«an

an

vow

3083.3. 62
3.52.— ...muam 3:3.— ofl...“ 55

26.590
2 25......
message

Es. his...
.. Na...
2...

z 6.590
z 62.92..

a use. 6.590

m
u>:uau:=nre=_

._o=a~_

a...

.::om

 

 
 

0.: a. 89:3...— w.._._._e..u 3:33.35... 183.8 .3 88:0 395.... :8 1... 23.. 09:02. 5.3.9.... 32.355

3.» 03:.

82

3. Method and Data
3.1 Method:

To address the aforementioned policy question, linear programming (LP) is used to
solve a farm household model. Since time is a key factor here to capture changes in soil
fertility over time and how they are valued according to the discount rate, multiperiod LP can
be used (Bafl‘oe et al., 1987).

As said earlier (figure 1), the LP model uses the simulated yield and soil nutrient
impacts presented above as model coeflicients. Treating soil fertility impacts as parameters
in optimization models is one of the approaches used in the literature to do a balanced
economic and environmental analysis of alternative systems (Roberts and Swinton, 1996;
Teague et al., 1995).

Bioeconomic models that link biophysical simulation to intertemporal optimization
models have been increasingly used during the last decade to measure and compare the efl‘ects
of farming practices: crop rotation, technologies, management decisions, etc. (Oriade and
Dillon, 1997). An early example is provided by Baffoe et al., (1987) who used multiperiod
LP techniques and the Universal Soil Loss Equation (USLE) to determine how several
representative crop rotational systems compare with each other and with monocultural corn
systems in Ontario from an economic and a land degradation (soil erosion and subsequent
efl‘ects on productivity) point of view. More recently, Barbier (1996) used a combination of
a recursive LP and an Erosion Productivity Impact Calculator (EPIC) model of soil conditions
and plant growth for two different agroclimatic zones of Burkina Faso under various

(population, market, prices, soil fertility) assumptions to test Boserup‘s hypothesis of the

83

effect of population pressure on agricultural intensification. Dalton (1996) adapted EPIC to
the Southern Mali context and linked it to a farm household model to study the long run
impacts of technical change (improved cultivars, crop residue management, organic
fertilization) and of policy alternatives (taxes) on crop production and land degradation.

In our work, like Baffoe et al., (1987), we compare different crop management
practices of a millet-peanut rotation from the profitability and soil fertility point of view, but
use DSSAT instead of USLE for the land quality impact estimation. Second, our analysis
differs from the ones above on (1) the type of crop simulation model used (DSSAT instead
of EPIC), (b) the inclusion of nonfarm activities as choice variables for the farm household,
and of (c) a soil capital reserves shock to mimic the effects of the phosphate distribution
program in Senegal.

Choice of DSSAT in our research has been justified by two advantages it holds over
other crop grth simulation models. First, it provides a‘ more detailed accounting of
phenological development and stresses encountered in each phenological stage than other
models (Kiniry, 1991; Jones et al., 1991), hence enabling a more accurate prediction of
variation in crop yields from year to year under different planting dates. Second, it requires
only moderate amounts of input data, and thus is relatively user-fiiendly. Moreover, initially
built with a nitrogen focus, the addition of a phosphorus component has made it suitable for
our purpose of simulating both N and P dynamics in a country with important rock phosphate

deposits and where P-deficiency in soils is said to seriously limit agricultural production.

84

3.2 Data:

Our main source of household data is the ISRA/IFPRI baseline data set collected from
a sample of 140 rural households in different zones of the Peanut Basin of Senegal during the
October 1988 - December 1991 period. It includes production, income and expenditure
variables. Added to that is another more recent data set collected under the 1995/96
ISRA/PRISAS/MSU single-visit farm survey and which covers basic characteristics (resource
endowments, crop mix, input use) of a 120-household sample from the same zones of the
Peanut Basin. Both data sets are used to determine the typical characteristics of farm

households in the two study zones (table 3.2).

Table 3.2: Selected characteristics of the typical farm household in the Senegalese
Peanut Basin

Characteristics Center Peanut Basin South Peanut Basin
Farm size (ha) 8 1 1

Labor force (man equivalent) 4 4

Ag equipment: animal traction (#) 1 hoe, seeder, horse 1 hoe, seeder, horse
Share of non farm income in total 24 29
household income (%)

Annual income (CFA/AB) 56000 72000

Source: ISRA/IFPRI and ISRA/MSU surveys (1988-92; 1996).

Partial crop budgets for millet and peanut under different cropping practices are

calculated with data collected in 1997/98 from ISRA (Senegal Agricultural Research

85

Institute) and CSA (National Grain Market Information System) grain price series of millet
and peanut in the Peanut Basin and input prices for seed, fertilizer, urea, etc. Production

coefiicients (human and animal labor needs, seed, etc) are obtained from Martin (1991)”.

4. Empirical Farm Household Model
The farm household model is mathematically specified as follows:
Mu El U(Y.)l = 2. (1+5)4 E {Y.}

= 2t 2| E] 21: (1+6)4 E{pjt*(Ajt* mu)“: ' rt: * ajlt + war . nut ' (l+i)*Cl‘.} (1)

subject to:
2, (am * A”) s b“ (2)
2,. A“ NB” 2 0 (3)
2.2. rm. * q...) m 2 cc. (4)
Lrv,-aE[Y,] $0 (5)
w..'n_.-BE[Y.]50 (6)
Dwn.-yCr. =0 (7)
am, .., Cr, 2 0 (8)

where

Y. is the net farm returns to crop and off-farm activities in the t period;

A“ is the level of the jth crop activity in the t period;

 

4 ISRA agronomists were also consulted to check the validity of these coefficients: no
major changes were detected, or if any, no empirical data are currently available.

86

q,“ is the average yield of the jth crop activity under the kth state of nature in the t

Pefiod;

a“. is the quantity of the ith resource used per unit of the jth crop activity in the t

Period;

p,” r. are respectively the output and input price vectors in the t period;

it... is the amount of labor devoted to nonfarm activities in the t period;

W... is the net returns to a day of nonfarm activity in the t period;

i is the interest rate;

Crt is the amount of credit received at time t;

a is the discount rate;

I).t is the endowment level of the ith resource in the t period;

NB” is the average annual changes in the level of soil nutrients for a hectare of land

devoted to the jth crop activity;

P reads probability of occurrence of the kth state of nature;

CC, are the household food grain requirements in the t period;

a, B, y are percentage values between 0 and l;

LIV. are the household total living expenditures in the t period; and

Dwn, is the amount of downpayment required on the received credit in the t period.

The model is constructed for the typical household in each of the two zones covered
in this study (Center and Southeast of Peanut Basin). These two zones are selected because

they are the main producers of the most important cash crop (namely peanut) in the country.

87

Also, spatially, they represent two agroclimatically different zones which are expected to
induce difl‘erential land quality and productivity impacts of cropping practices.
4.1 Activities:

Equation (1) contains the different activities to be undertaken by the farm household
to maximize the objective fimction. Major activities are crop production and nonfarm
activities, input credit borrowing and grain purchases if necessary to cover food needs. Crop
production activities in the LP model are the millet-peanut cropping practices described in the
previous section. Nonfarm activities are also incorporated in the model. In addition, provision
is made for buying inputs on short-term formal credits to be reimbursed at the end of the
period during which they are contracted.

4.2. Constraints:

Constraints are shown in equations (2) to (8). Per period physical constraints are
placed on the human and animal labor, on cultivated land (equation 2) and on soil nutrients
(equation 3). Labor is supplied entirely fi'om the household to carry out on and off-farm
activities during different sub-periods (seasons) of the year, and there is no hiring of outside
labor.

Financial constraints are imposed on starting capital, credit available and amount of
nonfarm income using empirical observations in the study zones. The credit market in the
Peanut Basin is active as it involves most farm households as borrowers and/or lenders; it is
also segmented: sources for production loans are formal while consumption loans are informal
(Warning and Sadoulet, 1998). Consequently, only formal production credit usually for

peanut cropping is included here and is limited by the required downpayment (equation 7).

88

Limited employment opportunities in rural areas constrain nonfarm income earnings. To
reflect this, we use empirical estimates of the share of nonfarm in total household income in
the two difl‘erent zones (Fall, 1991) to put an upper bound on to how much nonfarm income
can be earned (equation 6). Changes in both financial and soil nutrient capital resources across
periods are monitored in the model.

For food security, grain consumption requirements must be satisfied by own
production and/or purchases by the household. The ‘safety-first’ model is chosen to specify
risk that is present in farming decisions made in this uncertain environment described earlier.
This simple and general risk specification is chosen over alternative ones for mainly two
reasons. First, the biophysical modeling of crop production used here already captures most
if not all production risks. Second, we did not find that more sophisticated specification of
risk (e. g., a chance-constrained model) gave satisfactory results, given the nature of the
simulated data produced by the biophysical model.

The failure to cover food grain needs because of production shortfalls is a common
won'y for a subsistence risk averse farmer. To shield the household against the risk of
insufficient coverage of food needs following production downfall due to weather, hence yield
variability, a ‘safety first’ constraint is set to allow the food security objective to be achieved
under difl‘erent states of nature (equation 4). Using the simulated yield results, three states
of nature (‘ba ’, ‘average’ and ‘ good’) are defined for each crop and level of fertilization on
the basis of the mean and standard deviation of the yield distribution (table 3 .3). Afterwards,
their corresponding probabilities are calculated. Thus, equation (4) ensures that expected

production plus purchases if necessary meet grain needs under all states of nature.

89

 

Table 3.3: Simulated mean millet and geanut yields (kg—[gag b; state of nature

Millet Peanut
Zone State of
nature No Semi- Intensive No Semi- Intensive
Center ‘Bad’ 197 425 492 237 563 754
‘Average’ 287 729 885 309 687 982
‘Good' 747 1041 1139 749 1089 1339
South ‘Bad’ 206 443 524 434 970 1359
‘Average’ 213 741 962 488 1098 1585
‘ ’ 767 166 3 949 7 00

Source: DSSAT3.5P simulation outputs.

In equation (5), living expenditures for other foods and needs are allowed to vary
positively with earned income. This specification makes total consumption expenditures
endogenous to farm income, which is consistent with economic theory (Adesina et al., 1988).
One could argue about including a minimum level of living expenditures, regardless of income
levels. We found in the analysis that this minimum level was always below the living
expenditures given by the model.

4.3. Right-hand side (RHS) values:

Average cultivated land is 8 ha/year in the center and 11 ha/year in the south.
Household size is respectively .10 and 11 adult equivalents in the two zones. Initial soil
nutrient stocks for N and P are set to observed levels in collected soil data sets: 34 and 26
kg/ha for labile P, 71 and 70 kg/ha for inorganic active P, 86 and 104 kg/ha for organic active
P, 33 and 30 kg/ha for inorganic N and 3333 and 6348 kg/ha for organic N respectively in the

center and the south. Starting cash capital amounts to respectively CFAF 62,500 and 75,000

90

 

in the two zones. Grain consumption requirements for the whole household are calculated on
the basis of the national norms of 185 kg of cereals per capita and per year. A summary of the
corresponding LP matrix is presented in table 3.4.

4.4. Other assumptions and scenarios:

The model covers 5 periods, each 2 years long (because of the two-year millet peanut
rotation), hence a total of 10 years. Other assumptions are: based on trends from historical
input and output price data from 1977 to 1996, product price and returns to nonfarm are
allowed to increase by 6% per period while input costs increase by 8% per period. Living
expenditures (exclusive of grain consumption, handled by equation 5) amount to 80% (a=.8)
of net household cash income which is farm revenues net of input costs, grain purchases and
credit repayment and the balance of income is transferred to the next period. The share of
nonfarm income in total income, [3, is initially set respectively at .25 in the center and .3 in
the south. The interest rate is set at 10% and downpayment on formal credit is 20% (v =2).

Different scenarios of the LP model are run, each representing an observed or
hypothetical policy situation under which the typical farm household operates. Alternative
situations range from the less restrictive conditions of the current policy program (phosphate
program, access to credit and nonfarm activities, called scenario A) to a very restricted access
to cash income (no credit or nonfarm; scenario D) through intermediate situations (no
participation in the phosphate distribution program or scenario B; either credit or nonfarm
being available or scenario C), all scenarios being run with and without a soil replenishment

requirement (i.e ending levels of nutrient stocks being higher their beginning levels).

91

.o e. c 30...; a. .32 3.5.... .... ..o...» 6.3 .595... ”z ”3.3%.... x. .8296“. 2.. 3.2.0.. 5...» =3": .9... .. .83.. - 5.62.0... 0.. :33.
as. .8... 8838... 85828 H... 5...... ..e. 3.3:... .2me as... 328.8 ...8... m .8295. N .83. . 222 .o as. .o 3522.. ”2....
3.68.. ..o 8.... 95.8 “E... 30...... ..e 8.... m...=8 E... 20...... ..e 8.... 39.8.... “.5... ...euuoa.
£552.. .5. o... E u... 5.. .038 5...... .2. E v0.3.0... .9. 0... 958.250 .2... 8...»..8 805.356.. v.8 5&5: an... 0.98 8.60. o. 62.2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.. .. u 2 E8 2
_. c u 2 Es ..
.. a n 2.8 a...
n... .. m .9.» M5......
. . .. .7; E... a... c - 38.x...
. 2.... .. m as... 55....
_ _. . .. N as... .8... .5
_ _. _ .. N «as. .8... .5
. .- . o N .3... .8... .6
8. .. . .. N as... 88......
8. .. . .. N «a... £98....
8. .. . .. N .35. 698.6
. 8an 88.. see
. . . m. a. a .. w .83 5....

. a w .2:. ..

_ 3.32 Sam... .. m a as. ..
. .28.. a... 8...... .. m . as. 558

. 8. 82s.... .. m n as. ..

_ 33:8 .93.... .. m . as. ..
. 858...»... .. w .033...
..uooo ..o. .31. 0.3. 5.388.. 8.1..
N 1...... . 2. 2...
. 3.32 :58. on? .8: 21......
. 8.5.8:... 3..- .8... 1......
. .. .. 338.3%: .. . .1538...
. a... 28. m ..8 a. .5.

2 ...8 5.. ma. .268: mu. .11..
2 2 EYE. .98 a. :20 .2... a. .35. .2:... use .3... s... 83. 8.6:.
Ezfggiuggiiigseoszg 3&2“. migfizoo
8.225..
3:... 3...... 2.... 3...... 5...... .3 o... .... Eafifiam "can 0......

 

92

5. Results

Table 3.5 summarizes the results of the LP model where only optimal cropping
practices under each of the four scenarios are presented. The following salient points can be
made.

(i). No traditional i.e non-fertilizer-using practices with and without recommended
seeding densities for peanut are present in the lO-year optimal plan, under any scenario. In
contrast, in all periods, all scenarios and all zones, the adoption of fertilizer-based practices
is optimal, given the farm household’s resource constraints and food security needs.

(ii). The optimal land allocation shows a diversified set of cropping practices that
include different schemes of fertilizer use: full intensification of millet-peanut under the
phosphate program (MPZPP), semi-intensification (MP-l l) and full intensification on peanut
only followed by non-fertilized millet (MP-02).

(iii). The full intensification scheme (MP-22) enters the optimal plan only under
scenario B, i.e outside the phosphate program. Under this scenario, the optimal solution in
the Center indicates a shifting from allocating land equally to semi-intensified (MP-1 l) and
intensified millet-peanut (MP-22) in the first period to cropping more land with intensified
millet- peanut than with the semi-intensive practice in the last four periods. Contrastingly, in
the South, the shifi (in terms of land allocation) from rotating firlly fertilized peanut with non-
fertilized millet toward semi and full intensification of both crops is also taking place.

(iv). Land and capital resources are used up under all scenarios in the Center whereas
family labor is binding only during weeding times. As expected, there is some idle labor when

nonfarm activities are not available or accessible to the household (scenario C). Land and

93

Table 3.5: Optimal long-term cropping plan (in ha/year) for the typical farm

household in the Senegalese Peanut Basin under different scenarios.

A: credit, NF, PP Bzcredit, NF, no PP C: Credit and PP D: No credit, no NF"

W

Period 1

MP-ll .l 2.4 3.9 2.3 .1 3.4 - -
NIP-22 - - .05 - - - - -
MP-02 6.2 5.8 4.0 5.8 6.2 5.1 2.5 3.2
MPZPP 1.7 - - - 1.7 - - -
Period 2

MP-ll 1.8 4.3 2.8 4.4 1.8 4.3 - -
MP-22 - - 4.8 3.5 - - - «-
MP-02 1.4 .5 .4 ~ 1.4 .5 1.9 3.8
MPZPP 4.8 3.6 - - 4.8 3.6 - -
Period 3

MP-ll 1.8 4.3 2.8 4.4 1.8 4.3 - -
MP-22 - - 4.8 3.5 - - - -
MP-02 1.4 .5 .4 - 1.4 .5 1.4 4.4
MP2PP 4.8 3.6 - - 4.8 3.6 - -
Period 4

MP-ll 1.8 4.3 2.8 4.4 1.8 4.3 - -
NIP-22 - - 4.8 3.5 - - - -
MP-02 1.4 .5 .4 - 1.4 .5 1.1 5.0
MP2PP 4.8 3.6 - - 4.8 3.6 - -
Period 5

MP-ll 1.8 4.3 2.8 4.4 1.8 4.3 - «-
MP-22 - - 4.8 3.5 - - - -
MP-02 1.4 .5 .4 - 1.4 .5 ~ .2 .8
MP2PP 4.8 3.6 - - 4.8 3.6 -

Nutrients

Lab P chg. l8 5 16 3.3 18 5 1.7 -.5
In. P. chg. -19 —15 -19 -15 -19 -15 -3.5 -7
Org P. chg -30 -28 -30 -27 -30 -28 -5.5 -12
In. N chg. 109 45 116 43 109 45 7.8 9
Org N chg -327 -597 -333 -569 -327 -602 -62 -250
Objective

function: 259 685 177 573 179 493 24 183

%

Notes: MP] 1 menu-intensification on millet-peanut rotation; MP22: full intensification; MPOZ: full
intensification on peanut only; MP2PP: full intensification under phosphate program.

Lab P, In P., Org P. (active pool), In N and Org N chg are total changes in the stocks of nutrients
in different pools (labile and active for P, inorganic and organic for N) at the end of the lO-year period and
are in kg/ha; positive (negative) numbers indicate build up (depletion) of soil nutrients.

" Results for this scenario are obtained when food security requirements are dropped in both zones.

94

labor constraints are not binding under the very restrictive (no credit or nonfarm) D-scenario
In the South, it is worth noting that land is never binding under any scenario: there are around
2-3 ha of land lefi unused by the model, as a result of the capital constraint, the only resource
that is always binding in this zone.

(v). Under the D scenario, which reflects the situation of the poorest of the poor
farmers, meeting grain consumption needs is unfeasible under initial resource constraints. The
consumption constraint was then dropped to allow a feasible solution to the model. Also,
most of the available land is unused in the last period of the optimal plan, because of the lack
of surplus income made in previous periods and carried over to following periods to finance
farm production. Millet production under this scenario varies over the different periods
between 2 and 36% of food grain needs in the Center and 11 and 67% in the South.

(vi). Removing nonfarm activity from the model to leave only credit (scenario C) has
almost no effects on the optimal hectarage plan compared to scenario A in both zones, the
only difference being observed in the first period in the south. This similar land allocation
stems from the fact that in A, all labor is used up for farm first and the rest for nonfarm. Thus,
removing the possibility of using labor for nonfarm does not change the optimal land
allocation, but only decreases the level of cash income made and also increases the amount
of credit needed. Overall, the most binding constraint is that of initial capital which conditions
the path (based on optimal cropping practices) to be taken in later periods. ~

(vii). Sensitivity analyses were done on all scenarios but D. Capital constraints were
eased by reducing credit downpayment requirements fi'om 20% to 12.5%. This change

allowed farmers to afford early investments in capital-demanding cropping practices such as

95

MP-22; more interesting, credit needs can even be lessened in later periods because of higher
cash income net of production costs and credit reimbursement being carried over to the next
period to finance input acquisition.

(viii). The effect of these optimal cropping practices is, under all scenarios, a
replenishment or, even better, a build up of the two plant-available soil nutrient stocks after
ten years. Labile P and inorganic N are being replenished in both zones, the highest build up
being reached under the A and C scenarios. On the contrary, the stocks of non-directly
available nutrients (inorganic and organic P in the active pool, organic N) are being depleted
under all scenarios. This stems fiom the flows between the different nutrient pools and fi'om
the mineralization process that releases inorganic forms of these nutrients that the crop plant
can use.

(ix). Lastly, in terms of overall profitability, the A scenario with credit, nonfarm and
phosphate program yields the highest discounted net income level aficr ten years, followed
by B (no phosphate program) and C (no-nonfarm, credit-only). These income levels are
always higher in the south than in the center of the Peanut Basin, because of higher yields and

more land being cropped.

6. Conclusions

The LP modeling results show that the ‘Agricultural Program’ policy measures, if
maintained, offer good incentives to push farm households in the Senegalese Peanut Basin
towards an intensification of millet and peanut production. Such an intensification path is

made possible by the farm profitability of fertilizer use. The positive impact on soil fertility

96

is illustrated by the build up of the plant-available soil macronutrients, namely inorganic N
and labile P and the slower depletion of the other nutrient pools. By enabling soil nutrient
replenishment, fertilizer use helps prevent future production losses or maintain future
production flows, controlling for other factors. These positive biological (production),
economic (profitability) and ecological (soil nutrient replenishment) impacts of optimal
cropping practices contribute altogether to making this intensification path sustainable.
Moreover, they provide needed evidence confirming that increased fertilizer use is a key
element in establishing a sustainable intensification of agricultural production in semi arid
areas of SSA. For example, one can recall from table 3.1 that not using fertilizer at all on
millet peanut rotation leads on average to a depletion of the plant-available labile P pool by
1.5 kg/ha/year in both studied zones. In contrast, the LP solution in table 3.5 suggests a
combination of fertilizer-based practices that replenish the same P pool under the A scenario
by 1.8 kg/ha/year in the Center and by .5 kg/ha/year in the South. .

Hence, results suggest some interesting prospects for promoting policies to ensure
easier availability and accessibility to this input. One policy implication of these results is the
necessity to ease initial capital constraints. Within the context of our model, cash and formal
credit constraints, especially in the initial periods, drive the crop production intensification
process. Results suggest that measures such as reducing downpayment requirements (or other
equivalent ones that would expand input credits or improve access to them) have a potential
for helping capital-deprived farmers make the necessary investments in productivity-enhancing
and soil nutrient-replenishing inputs. But, this model is restrictive in many senses (formal

credit only, no migratory income or other income transfers, etc). Thus, a question is how well

97

it is reflecting the reality of credit constraints in rural Senegal. How really binding these
constraints are constitutes a research issue that needs to be empirically established.
Another implication for policy is that programs to remedy the P-deficiency of soils
through the distribution of phosphate products can be important because of their production
efl‘ects compared to currently observed no-fertilizer using practices. However, for them to
have any chance to reach the production and nutrient replenishment objectives, the necessary
accompanying conditions must be satisfied, i.e annual fertilizer use by farmers, and timely and
correct incorporation of P-products in the soil. Soil fertility management is complementary
to soil amendments. Consequently, further extension and monitoring efforts should be
deployed to inform and convince farmers of the need to correctly undertake these
complementary actions. If not, leakage instances like the resale of P-products or their use
for other purposes (e.g. construction bricks), their inappropriate application on the soil
surface and after the first rains will be more common, and will place this program among the
numerous theoretically sound but unfortunately ill-implemented agricultural policy programs.
Finally, this study did not incorporate organic fertilizer, mainly because of data
limitation problems. Despite that, it should be understood, as mentioned in the general
introduction, that a combination of both organic and inorganic fertilizers is necessary to
achieve land productivity and quality goals. Hence, further research efforts should include

both forms of fertilization in biophysical and economic modeling.

98

References

Adesina, AA, P.C. Abbott and J .H. Sanders. 1988. Ex-ante Risk Programming Appraisal
of New Agricultural Technology: Experiment Station Fertilizer Recommendations in
Southern Niger. Agricultural Systems, 27(1988):23-3S.

Bafl‘oe, J.K., D.P. Stonehouse and ED. Kay. 1987. Methodology for Farm-level Economic
Analysis of Soil Erosion Effects under Alternative Crop Rotational Systems.
Canadian Journal of Agricultural Economics 35:55-74.

Barbier, B. 1996. "Impacts of Market and Population Pressure on Production, Incomes and
Natural Resources in the Dryland Savannas of West Afiica: Bioeconomic Modeling
at the Village Level" Intl. Food Policy Research Institute: EPTD Discussion Paper
No. 21.

Binswanger, H. 1990. Attitudes Toward Risk: Experimental Measurement in Rural India.
American Journal of Agricultural Economics 62:395-407.

Cleaver, K. and G. Schreiber. 1992. The Population, Agriculture and Environment Nexus in
Sub-Saharan Afiica. Agriculture and Rural Development Series No. 1, Technical
Department, Afiican Region, World Bank, Washington, DC.

Dalton, T.. 1996. Soil Degradation and Technical Change in Southern Mali. Ph.D
Dissertation. Purdue University, Department of Agricultural Economics, West
Lafayette, IN.

Diagana, B., V. Kelly and M. Kébé. 1996. L’Offre Agricole Suite a la Devaluation: Pourquoi
Une Réponse Si Faible Au Séne'gal. ISRA/PASE-CILSS/PRISAS Note, October.

Delgado, C., J. Hopkins and V. Kelly. 1994. Agricultural growth linkages in sub—Saharan
Africa: A synthesis. p. 22-26. In Proc. of a Workshop on Agricultural Growth
Linkages in Sub-Saharan Afiica, Washington, DC. 26 May 1994. Int. Food Policy
Res. Inst, Washington, DC.

Fall, AA. 1991. Composition Multisectorielle et Distribution du Revenu du Menage Rural:
Cout d’Opportunite, Remuneration des Facteurs Capital et Travail. ISRA/IFPRI
Project Document V, January. ‘

Freud, C., E. H. Freud, J. Richard and P. Therein. 1997. "La Crise de l’Arachide Au Senegal:
Un Bilan-Diagnostic" CmAD report, January.

99

Gaye M., M. Sene and V. Kelly. 1996. Facteurs Determinants des Densites de Semis de
l’Arachide. Michigan State University Department of Agricultural Economics Staff
Paper no. 96-29.

Hanks, J. and J. T. Ritchie (eds). 1991. Modeling Plant and Soil Systems. Number 31 in the
series of Agronomy. American Society of Agronomy, Inc. Madison, WI, 71-90.

Jones, C.A., K.J. Boote, S.S. Jagtap, and J.W. Mishoe. 1991. “Soybean Development” Ch.
5 In J. Hanks and J. T. Ritchie (eds). Modeling Plant and Soil Systems. Number 31
in the series of Agronomy. American Society of Agronomy, Inc. Madison, WI, 71 -90.

Kelly, V., T. Reardon, B. Diagana, A. Fall and L. Mcneilly. 1993. "Final Report for the
IFPRI/ISRA Study of Consumption and Supply Impacts of Agricultural Price Policies
in the Peanut Basin and Senegal Oriental" (2 volumes), submitted to USAID/ Senegal,
September.

Kiniry, JR. 1991. “Maize Phasic Development”. Ch. 4 In J. Hanks and J. T. Ritchie (eds).
Modeling Plant and Soil Systems. Number 31 in the series of Agronomy. American
Society of Agronomy, Inc. Madison, Wisconsin. 1991: 55-70.

Larson, BA and GB. Frisvold. 1996. Fertilizers to Support Agricultural Development in
Sub-Saharan Africa: What is needed and Why? Food Policy, vol. 21, no. 6: $09-$25.

Marter, A and A. Gordon. 1996. Emerging Issues Confronting the Renewable Natural
Resources Sector in SSA. Food Policy 21(2), pp. 229-241, May.

Martin, F. 1991. Budgets de Culture au Senegal. ISRA/Michigan State University,
Department of Agricultural Economics, vol. 4, no. 5.

Mbodj, M. 1987. L’Utilisation des Engrais et la Production Agricole: Fertilisation des
Cultures au Senegal et dans les Autres Pays de I’Afiique de l’Ouest. Mimeo, Ministry
of Rural Development, Dakar, Senegal.

Mudahar, M. S. 1986. Fertilizer Problems and Policies in sub-Saharan Afiica. InManagement
of Nitrogen and Phosphorus Fertilizers in sub-Saharan Africa. eds A. U. Mokwunye
and P. L.G. Vlek Proceedings of a symposium. Martinus Nijhoff Publishers.

Oriade, C.A., and CR. Dillon. 1997. Developments in Biophysical and Bioeconomic
Simulation of Agricultural Systems: A Review. Agricultural Economics. 17: 45-58.

Padwick, G.W. 1983. The Maintenance of Soil Fertility in Tropical Africa: A Review.
Experimental Agriculture. vol. 19: 293-310.

100

Reardon, T., C. Barrett, V. Kelly and K. Savadogo. 1999. Policy Reforms and Sustainable
Agricultural Intensification in Africa. Development Policy Review, forthcoming.

Reardon, T., V. Kelly, E. Crawford, B. Diagana, J. Dione, K. Savadogo and D. Boughton.
1997. Promoting Sustainable Intensification and Productivity Grth in Sahel
Agriculture after Macroeconomic Policy Reform. Food Policy, 22(4): 317-327.

Roberts, W. S. and SM. Swinton. 1996. Economic Methods for Comparing Alternative Crop
Production Systems: A Review of Literature. Reprint from American Journal of
Alternative Agriculture, vol] 1, no. 1.

Shapiro, B. and J. Sanders. 1998. Fertilizer Use in Semiarid West Afiica: Profitability and
Supporting Policy. Agricultural Systems, 56(1998), no.4.

Senegal, Republic of, Ministry of Agriculture. 1996. Programme Agricole 1997/98.
December.

Senegal, Republic of, Ministry of Agriculture, Bureau Pédologie. 1997. Programme de
Restauration de la Fertilité des Sols du Séne'gal. April.

Senegal, Republic of, Ministry of Agriculture, Bureau Pédologie. 1995. Rapport Annuel.

Senegal, Republic of, ICS/SENCHIM. 1996. Projet de Restauration de la Productivite des
Sols au Senegal. Mimeo, October.

Sheldon, V. 1987. Fertility Status of Afiican Soils: a Handbook for Non Agronornists.
Curriculum Publications Clearinghouse, Western Illinois University.

Teague, M.L., D.J. Bernardo and HP. Mapp. 1995. Farm-Level Economic Analysis
Incorporating Stochastic Environmental Risk Assessment. American Journal of
Agricultural Economics. 77(1995): 8-19.

Warning, M. and E. Sadoulet. 1998. The Performance of Village Intermediaries in Rural

Credit Delivery under Changing Penalty Regimes: Evidence from Senegal. Journal
of Development Studies, 35(1), October, pp] 15-138.

10]

"unounurnnurirrr