V'w‘f

V—

 

 

 

 

 

.
.
.
.
‘ .
.
c
. ‘
.
.
‘
.
.
.
.
.
.
0.
~ . .
‘ T
.
.‘
.:
. .
o
0.
.
:
.
o
l
V O
.
I
.n .
n
.
‘ I
' v
.
.
~ a
I .
n
.
- ‘ ,.
. ‘ - .
4'
. .
' n
..
y l
.
- _ .
n
.
. C
| I
n
.. ..
5 .
. .
‘D c
.. ,
- .
. 4 .
v
_, .

__———— '__ — 7 v
- $.nL-.o u- .. . . ' _-‘».,-_m‘ __ .'.. ‘-

 

 

 

     

 

" “Oi-“v . 9- ,...

 

u I .' .. ‘ .. -
. . ‘ o
: ' g, .
‘ . .D
O
'l . - " {-1.
. . ‘ o _ ‘.
.. ‘ .
. ' ' .'..' ‘.I .
‘ . E..IM ,‘
. I.‘ " ~ I .2
'. I. ' . R RAM.-
‘ ‘ I ,, " ~ I. -
. ' . ‘ ‘ V . - ..‘ . ‘,. '_".
" . _ u._. _‘.
.' M ‘ .- ‘ . ,V ‘ I, \.. . I.
I. ' . ' .- 2'. " O- ' —“ .
' ' ‘ .‘ I ‘ . ".
. . . ',. _ v“ I;
, - o -- ’ ‘ . ' . . ¥ . ... ‘1 '
. _ . . ' . I
- - ' ~ ~ . - an. .
.. '

GNTHE

 

"W'er'x: .. __

 

 

   
     

      
     

  

ommomar

   

     
  
   

 

   

   
  
  
  
 

 

   

     

pr mm ‘ of

 

 

m
JAN 2 6 2000

MICH STATE UNIV

 

‘ LIBRARY
Michigan State

 

University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record. kit}, 86’
TO AVOID FINES return on or before date due. 6‘) ("9
MAY BE RECALLED with earlier due date if requested. ”72% ya
( .
DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CiFIC/DateDue.965-p.15

MICH. STATE UNIV

OENCU

 

The Impact of the Fertilizer Sub-Sector Reform Program on the Demand for Fertilizer

in“0fficedu Niger",Mali

Youssouf Cissé

A Plan B Paper

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

AIAST ER OF SCIENCE

Department of Agricultural Economics

1998

ABSTRACT

ThelmpaaoftheFerfilherSub-SectorkefoumgnmontheDemandforFatiizer

i. “Ollicedu Niger”,Mali

By

Youssouf Cissé

The “Office du Niger” (0N) is the largest irrigated rice production area in Mali.
In the ON, the use of chemical fertilizers is recommended in the cultivation of improved
varieties of rice in all production systems.

Given the importance of chemical fertilizer in rice production in the ON,
surpr'uingly little is known about the impact of the fertilizer sub-sector reform program
on the demand for fertilizers and the capacity of the ON to support a private-sector
fertfl'uer distribution system. Insights into these questions will ultimately help in
assessing the potential for a vibrant and sustainable chemical fertilizer subsector in the
ON. The primary objective of this paper is to describe how to carry out a study to assess
the impact of the fertilizer reform program on the demand for fertilizer in the Office du
Niger. It is planned that data will be collected from both primary and secondary sources
in order to implement the proposed models and their extensions. Primary data will he
collected from a stratified sample of 30 farms in each rice production system in the ON.

Secondary data will he collected by consulting relevant reports written about the ON.

3
Empirical models based on the conventional linear programming framework will

be bait to represent a typical household in each rice production system in the ON. The
models will be run using two types of price vectors: the input prices at the farm gate in
the ON before and after the devaluation of the FCFA currency. In each case, it is
possible to derive the aggregate potential demand for fertilizer by varying its price. As
the price of fertilizer varies, different levels of input use become optimal and, in
consequence, a sen'es of price-quantity relationships is developed. The risk of yield
variability in food crops and income variability from crop sales resulting from weather
and prices will be incorporated through use of the minimization of total absolute
deviation (MOTAD) model as developed by Hazel] in 1971.

The results from the analysis should provide insights about the incentives for
farmers to use chemical fertilizers and the capacity of the ON to support a private sector

fertil'uer distribution system.

This piece of work is dedicated to

my mother, Doussou Coulibaly,

Who never stood against my ambition to undertake further training,

and
my father, Abou Cissé,

who taught me that hard work and patience are the key to success in life.

ACKNOWLEDGEMENTS

Iwishtoexpnessmythanksand appreciationtomy majoradvhor,Prof.John
Staatz, who superv'ued this plan B paper, providing intellectual stimulation and
invaluable guidance. By reviewing the paper with patience he made useful suggestions
and constructive criticisms. I am also grateful to Prof. Eric Crawford, Prof. Steve
Hanson, Prof. Carl Liedholm for having accepted to be in my guidance committee.

I am grateful to USAID Mali for providing through the PARA project the
financial assistance, which enabled me to undertake further training.

I would like to express my thanks and appreciation to many people for their help,
'mtellectual and moral supports during my graduate program.

I would like to express my appreciation to all faculty and staff in the Department
of Agricultural Economics, Michigan State University.

Lastly but certainly not the least, I am indebted to all the members of my family
for their patience and sacrifice during my long absence for studies in the US. The smiles
of my three children , Abdrahamane, Abdoulaye and Ibrahima gave me one more

reason not to give up.

TABLE OF CONTENTS

ABSTRACT
DEDICATION
ACKNOWLEDGEMENTS
LIST OF TABLES
LIST OF FIGURES
LIST OF ABBREVIATIONS
CHAPTER I: INTRODUCTION
l-l-Problem Statement
l-z-Specific Issues
CHAPTER H: THE SETTING
2-l-Introduction
2-2- Market Reform in the Office du Niger
2-3- Fertil'ner Subsidy Removal
2-4— The CFA Franc Devaluation
CHAPTER HI: CONCEPTUAL APPROACH OF THE DETERMINANTS
OF FERTHJZER DEMAND
3-1- Demand of Fertilizer Under Perfect Market Conditions
3-2- Demand For Fertilizer Under Imperfect Market Conditions
3-3- Rule of Thumb: The VCR

3-4- Risk Considerations

N

10

ll

l2

l2

14

25

25

29

32

37

37

3-5- Financial and Economic Analyses 46
CHAPTER V: EMPIRICAL MODEL 50
4-1- Introduction 50
4-2- Review of Methods Employed For Estimation of Consumption 50
of Fertflizers
4-3- Mathematical Programming 54
4-4-Parametric Programming so
4-5- Empirical Model, Expect Results, and Modification of the Basic
Model 63
4-5-1- Empirical Model for the Study Area 63
4-5-2- Expected Results 71
4-5-3-Aggregate Potential Demand 71
4-5-4 Import Parity Price for Fertilizer 74

4-5-5- Modif'rcation of the Basic Model to Incorporate Risk Component 77

CHAPTER V: DATA NEEDS AND PROPOSED DATA COLLECTION

STRATEGY 85
5-1- - Data Needs 85
5-2- Data Collection Strategy 88
5-3- Sampling Procedure and Sample Size 88

APPENDIX A: Programming Matrix Including the Different Activities in the
ON Area 95

APPENDIX B: Programming Matrix Including the Possible Correlation Relationships

betweenRiceandotherEnterprisesintheONArea

BIBLIOGRAPHY

97

LIST OF TABLES

l- Fertilizer Prices in Current and Relative Terms 16
2- RelativePricesofFertilizerintheONinl994/95 l8
3- Other Computer Programming Techniques 58
4- Relative Importance of Vegetable Crop Production

(Percent of Households) 65
5- Ineorporation of Fertilizer Buying Activities 66
6- Incorporation of the Minimum Level of Income 81

7- Incorporation of the Minimum Level of Food Crops Self-Sufficiency 84

LIST OF FIGURES

1- Map of the Hydraulic Systems in the ON
2- Annual Prices of Paddy, Rice, Urea, and Ammonium Phosphate (AP)
Prices in Mali, 1992-97

3- Relative Prices of Urea and Ammonium Phosphate (AP), Mali, 1992-97

10

13

19

20

BNDA

CMDT

DNACOOP

DNSI

FAO

FDV

MOTAD

ON

ODR

PARA

SCAER

VCR

1 1
LIST OF ABBREVIATIONS

Banque Nationale de Developpement Agricole
(National Development Bank)

Compagnie Maliénne de Développement des Textiles
(Malian Company for the Development of textiles)

Direction Nationale de l’Action Cooperative (National Direction of
Cooperative Action), Mali

Direction N ationale de la Statistique et de L’ informatique.

Food and Agricultural Organization

Fond de Développement Villageois (Village development fund), Mah’
International Fertilizer Development Center

Linear Programming

Minimization of Total Absolute Deviation

Office du Niger (Niger River irrigated agricultural development agency)
Operation de Development Rural (Rural Development Agency), Mali
Programme d’appui a la recherche agronomique

Société de Credit Agricole et d’Equipement Rural (Agricultural credit
and input distribution agency), Mali

Value Cost Ratio

12
CHAPTER I - INTRODUCTION

l-l - Problem Statement

MaliisaeountrylocatedintheSahelianzoneofWestAfrica. TheMalian
economy is stfl dominated by the agricultural sector, which is characterized by its
reliance on erratic weather conditions and poor soils. Stoorvogel and Smaling (1990)
reportedthatMaliansoilslost8,2,and8kg/haofnitrogen(N),phosphonrs(P),and
potassium (K), respectively, during 1983. The projected annual losses for the year 2000
are 11, 4, 12 kg/ha for N, P, and K. Because of this situation, Malian soils require
nitrogen applications, and many require phosphorus, potassium, and sulfur applications
to maintain their nutrient balance and soil productivity (Henao et at, 1982).

Indeed, fertilizer use will play a special role in meeting the twin challenges of
sustaining food security and preserving the natural resource base. Fertilizer is
complementary to the use of seed, water, and plant protection materials capable of
shifting the production function and is essential for replenishing the nutrients removed
from the soils. The growth in fertilizer use should be sustained by the introduction of
policies and technologies to improve the efficiency of fertilizer use so that more output
can be obtained from the same or a lower amount of fertilizer nutrients (IFDC, 1995).
Only an efficient and environmentally sound fertilizer use can sustain high crop yields
and prevent the degradation of the resource base (IFDC, 1995)

On the other hand, if fertilizer use is not managed properly and if fertilizer is
used excessively, it can cause harm to the environment through nitrate leaching,

eutrophication, and other externalities.

13
In the “Office du Niger” (henceforth ON), which is the largest Malian irrigated

rice production area, the use of inorganic fertilizer (urea and ammonium phosphate) is
recommended in the cultivation of improved varieties of rice in the intensive (N iono),
semi-intensive (Koken’), and non fully restored production (Kouroumari) areas. Figure
1 provides the map ofthe hydraulic systems in the ON area. The intensive system enjoys
ful water control and involves the use ofa relatively high level offertilizer. The semi-
intensive system does not benefit from full water control in all areas of the field.
Therefore, farmers in the semi-intensive system use lower doses of fertilizer than in the
intensive system. The third system, called the non-restored area, is not maintained on a
regular basis, and farmers use a relatively lower dose of inputs. Kamuanga (1982)
reported from his study in the ON that 72% of the farmers used less than the
recommended level of 100 kg of fertilizer (urea and phosphate), while 18 percent applied
it at rates higher than 100 kg per hectare. Application rates reached the recommended
level of 50 kg of urea only in one zone (the Sahel), which also had the highest rate of
application for ammonium phosphate.

Given the importance of the chemical fertilizer in rice production at the ON,
surprisingly little is known about the impact of the fertilizer sub-sector reform program
on the demand for fertilizer in the rice production systems at the ON and the capacity
of the ON to support a private-sector fertilizer distribution system. Insights into these
questions will ultimately help in assessing the potential for a vibrant and sustainable
chemical fertilizer subsector in the ON.

Ekewhere in Africa, for example in Senegal, Kelly (1988) reported that the major

 

 

m<memme m<Um>cr_QCmm
c: cmrq> nmzam>r z_nmm_mz

Z>ZD>F>

 

xocxocz>m

x>r>
_zwmm_mcw

 

.:0rooo

 

as . T

. -I.
3>mx>r>,

  

gm 952. mm. 35m > no...» n>O>m.:~n 9..

g» .2)

xorozoosozo

 
    

Vfl%Q$10—.

 

 

           

ZOVAH

rmomzcm

Mmfipmc<m.fi>p>.
n>z>cx o._mx_o>e_oz
mHHNOZM >zmz>0mm

.222;
mH_m<mem=m =<om>cr.ocm

X>xmz _ m<m4m3m I<on>crHOCm

:IPn _ z m Nozm >2mzpmmm

 

 

14
reasons for the poor performance of the liberalization of the fertilizer market in the

1980s were the high risks and low payoffs of using fertilizer at unsubsid'med prices,
combinedwithcreditconstraints. Incontrast,insomepartsofsub—SaharanAfrica,
parficuhrltheI-watuedhighhndaitappunthatcomtraintsmexpandedueof
fertilizer and improved seeds lie mostly on the supply side (Kelly, 1988). Dione a at.
(1996) reported that the private sector in West Africa was still unable to satisfactorily fill
the gap resulting from the elimination of parastatal enterprises in charge of input
delivery before the structural adjustment reforms. The main constraints were as
folows:

1) Complexity of the input delivery function: Input delivery required

2) Satisfactory financial availability, knowledge of the world market and

also
3) Good technical knowledge of the inputs themselves.
4) Small size and poorly organized private sector: These characteristics
prevented private sector from benefiting from scale economies.
One of the key questions is which of the above constraints is most binding in
the ON 2
1-2- Specific Issues
Until 1980 in Mali, the agricultural credit and input distribution agency

(SCAER) was the monopoly parastatal in charge of the distribution of farm supplies and
the provkion of credit to farmers through the Rural Development Operations (ODRs).

SCAER was abolished in 1981 by the government and replaced by the National

15
Agricultural Development Bank (BNDA), which extended credit either directly to
farmers through guarantee of group liability of approved village associations or
indirectly through ODRs or other Rural Development Agencies, such as the National
Direction of Cooperatives (DNACOOP). Both SCAER and the BNDA operated in
relatively stable socio-economic and political environments. Today, however, there are
many forces leading to change in the Malian fertilizer subsector. The main forces

driving changes are:

i)The removal ofsubsidies

Fertil'ner subsidies in Mali used to be high in the mid-70s (more than 50% of the
actual fertilizer costs). The subsidies were intended to reduce the price of fertilizer, but
they were substantially reduced in the late 70s and early 80s (to a level of about 15% to
25%) until they were abolished in 1987. Table 1 provides the prices of fertilizer in

current and relative terms between 1970 and 1990.

16
Table 1: Fertilizer Prices in Current and Relative Terms

 

 

 

Year Urea Ammonium Price ratio
Phosphate
(CFAF/ks) (CFAF/ks)
Urea/ Urea/ Ammon.
Maize Paddy phosphJPaddL
1970/71 30 20 3.0 2.4 1.6
71/72 30 20 3.0 2.4 1.6
72/73 32 20 3.2 2.6 1.6
73/74 32 20 2.0 2.6 1.6
74/75 32 20 2.7 1.6 1.0
75/76 43 28 2.6 2.2 1.4
76/77 46 48 2.6 2.0 2.1
77/78 46 48 2.2 2.0 2.1
78/79 55 70 2.4 2.2 2.6
79/80 60 70 1.9 2.0 2.3
80/81 65 70 2.3 1.7 1.9
81/82 103 108 2.2 2.1 2.2
82/83 103 108 2.1 1.9 2.0
83/84 103 108 2.1 1.7 1.8
84/85 103 108 1.9 1.7 1.8
85/86 105 120 2.5 1.5 1.7
86/87 134 194 -- 1.9 2.8
87/88 145 194 -— 2.1 2.8
88/89 145 194 -- 2.1 2.8
89/90 145 194 -— 2.2 2.8

 

Source: adapted flan: Sijm (1992).

Table 1 provides the current prices of urea and ammonium phosphate in
CFAF/kg in the second and third columns, whereas the relative official prices
(fertilizer/cereal) are provided from the fourth to the last column. Since the suppression
of SCAER in 1981/82, the prices of fertilizer were no longer fixed by official decree but

announced, based on reference prices practiced by the cotton production agency CMDT

17
(for area) and the “Operation Haute Vallee” (for ammonium phosphate). It can be seen

from Table 1 that in absolute terms, the unit price of urea and ammonium phosphate
folowed an increasing trend from the 70s through the 90s. However in relative terms,
both the prices of urea and ammonium phosphate compared to cereal prices decreased
from the 70s to the mid-eighties before the complete removal of the subsidies. This
decreaseintherelativepriceoffertilizerinlate70’sandearly80’swasdueinpartto
high cereal prices resulting from drought (ag., in 1984/85) rather than explicit fertilizer
policy. After the removal of the subsidies, the relative prices increased. Then the CFA
devaluation took place on January 12, 1994, drawing more attention on the input and
output price relationships.

Dione a of. (1996) found that the price of inputs registered a 71% increase
between the 1993 and 1995 cropping seasons. They reported that many rice producers
in Mali responded to this trend by reducing the level of fertilizer use by 11% (Dione d
01., 1996). This changing input prices hurt farmers and village associations that could
not get access to formal credit because of their high level of indebtedness most. They
were forced to rely on the informal source of seasonal credit from traders where the rate
of interest ranged from 20% to 50% (Dione et al., 1996).

Table 2 below presents the prices of fertilizer relative to the prices of paddy in
each zone in the ON. It can be seen from table 2 that the relative prices were higher in

the more intensive areas implying that paddy prices were lower in these areas .

18
Table 2: Relative Prices of Fertilizer in the ON in 1994/95

 

 

Zone Absolute prices of Relative
1' IT P .
(CFAF/kg)
Intensive (Niono) Urea: 178 1.78
AP: 172 1.72
Semi-intensive Arpon (N iono) Urea: 178 1.66
AP“): 172 1.61
Semi-intensive Arpon (Macina) Urea: 165 1.37
AP : 186 1.55
Non-Restored area Urea: 165 1.35
AP: 178 1.46

 

('i The relative price is the price of fertilizer over that of paddy in each Zone.
0’ AP =- Ammonium Phosphate.

Source: Derivedfronr 0N raw data by the author.

Figure 2 presents the prices (CFAF/kg) of paddy and rice along with the prices of
urea and ammonium phosphate from 1992 to 1997. In general, there was an increasing
trend in the prices of both paddy and rice from 1992 to 1996. But from 1996 to 1997
both paddy and rice prices dropped. The decrease in the prices of rice was more drastic.
There was an increasing trend in the prices of both urea and ammonium phosphate.
The prices of both fertilizers increased drastically after the currency devaluation in 1994
untfl 1996. The decreasing trend observed in the prices of both fertilizers from 1996 to
1997 was visibly less pronounced than in the case of paddy and rice inrplying that

fertil'ner prices outweighed paddy and rice prices over the same period.

19
Fitz-Ire3

presentstheprices

g1:
... of urea and

I”
f“

     

   

1802 I.“

V.

 

[er-his: -o-nu +um +rs j .
Figure2: AnnualPricesofPaddy,Rice, Urea, rho-plate relative
and Ammonium Phosphate (AP) Pnces in Mali, 1992-97. to those of

 

Source: Derived fiom 0N raw data by the author
and milled rice. The

prices relative to

paddy drooped

from 1993 to 1995

implying , that

paddy price-

outweighed fertilizer prices over this period. But from 1995 to 1996, the inverse trend

was observed. The prices relative to paddy dropped slightly from 1996 to 1997. The

fertilizer prices relative to those of milled rice increased over the period 1994 to 1997

implying that fertilizer prices outweighed rice prices over that period. Since fertilizer is

a critical input in rice production in the ON, an increase in its relative prices rakes the
issues of input cost and use efficiency.

Indeed, the relative prices on both tables 1 and 2 need to be compared to the

marginal physical product or the marginal unit of rice produced from each unit of

fertfl'ner, bringing about an allocative efficiency issue. In other words, one needs to

compare the value of the rice produced with an additional unit of fertilizer (MVP of

20
faflher)withthefatfl'nerpficeinordertoidenfifytheefficiencywithwhich fertilizer

hmedbyfamanthepotenfialdemandforferfilizermdanyoppoflunityforcesu
reduction. In other words, only if the marginal value product of fertilizer exceeds its unit
cost will further purchases of this input be profitable and the most profitable level of
fertilizer will be that for which marginal value product offertilizer (MVPn) equals it
uniquecostP(Xn). Intermofdemand,thegraphofMVPnagainsttheuaeoan may
be interpreted as a farmer’s demand function for this input indicating the quantity that

he should purchase at various factor prices.

     

OJ

     

1.84 l... 1887

Th-

l-o-WM -o-rr/ruw +0m/Ivo +0/Ih- I
Figure 3 : Relative Prices of Urea and Ammonium Phosphate (AP),
Mali,1992-97.

 

 

Source: Derived fi'om 0N raw data by the author

21
ii) The liberalization of the input/output markets:

TheMalianricemarketiwas liberal'ned in 1986 and ricefarmers in theONwere
alowedforthefirsttimetoseltheirproduction in themarket. Priorto 1986,theywere
obliged to sel their entire marketable surplus, at the official price, to the ON rice mills.
In addition, the fertilizer market was liberalized in 1994, attracting more attention to the
input and output price relationships and stimulating the debate about input demand

andsupplyissues.

iii) The Devaluation of the CFAF currency:

On January 12, 1994, the exchange rate between the CFA franc and the French
francdroppedbyhalf,from 1FF=50CFAFtol FF=100 CFAF. Theargumentfor
devaluation is that changing the nominal exchange rate will increase incentives for
domestic production of tradable goods (exportable goods and import-substitutes) and
discourage domestic consumption of 'those goods (Staatz et al., 1994). This process '3
supposed to lead to re-establishment of external balance in foreign trade and greater
intersectoral balance in the domestic economy.

Sanders et a1. (1996) argued that the devaluation in 1994 will have a short-run
effect in reducing the demand for imports of inorganic fertilizer, but that a long-run
effect will be to increase prices of domestically produced food as imported agricultural
products become more expensive. However, to the author’s knowledge, no empirical
studies have been carried out on the impact of input/output market liberalization and

the 1994 currency devaluation on the fertilizer market in the ON. Malian economic

22
policy-makers and researchers, however, need to know the answers to two key questions.

First, how have market reforms affected the demand for and supply offertilizer in the
ON? The answer will depend in part on how these changes have affected the
profitability of producing rice and other commodities in the ON. Second, what actions
(3 any) should be implemented to improve the performance of the inorganic fertil’ner
subsector in the ON? The relevance of the foregoing questions is at three levels. First,
the ON has been the most important rural development agency in Mali with regard to
irrigated rice production, processing and marketing. Since the early 80s, the ON has
become part of a comprehensive reorganization process under the structural adjustment
programs in order to improve its performance, but particularly to increase the self-
reliance and socio-economic position of its rice farmers. It is believed that the reforms
within and outside the ON open the perspective that the country might become a
competitive and self-sufficient producer of rice in the 1990s (Sijm, 1992). Therefore, it
becomes important to carry out a study in order to answer some of the questions that
policy makers and researchers are concerned about.

Second, a study on demand for inorganic fertilizer can enhance the efficiency of
resource planning on the supply of fertilizer and related inputs. The knowledge about
the importance of factors affecting demand can enable the private sector to act not only
on the planning of supply but also on relaxing the factors that constrain the fertilizer
demand at the farmers’ level. Finally, policy makers interested in transforming
agriculture in general in Africa and specifically in Mali need empirical knowledge not

only at the macroeconomic level but also at the microeconomic level in order to see

23
whether the market reforms are performing well.

This paper outlines an approach to answering research questions about the
dflferent rice production systems in the ON:
(1) What is the level of the demand for fertilizer at the ON given current
And Possible future prices?
(3) At what level can ON support a private-sector fertilizer distribution

system?

These research questions will be answered against the working hypotheses stated
below. We have two working hypotheses related to the demand for fertilizer at the farm
level after the market reforms:

1) Effective demand for fertilizer still exists across rice production systems

in the ON given current prices
2) There exist a production and a fertilizer demand levels at which the ON

can support the private distribution system private sector

The primary objective of this paper is to describe how to carry out a study to
assess the impact of the fertilizer sub-sector reform program on the demand for fertfl’uer
in the Office du Niger. The specific objectives of such a study would be:

(1) To evaluate the potential demand for fertilizer by farmers in the ON rice
production systems,

(2) To assess the capacity of the ON to support a private sector fertilizer

24
distribution system.

(3) To provide from the results of the study insights about the incentives of
farmers to use fertilizer and the costs involved in supplying it. This
procedure will help evaluate the capacity of the ON to support a private
sector fertilizer distribution system and suggest any policy changes needed to

overcome problems.

25
CHAPTER II - THE SETTING

2-1- Introduction

This chapter reviews the effect ofstructural adjustment policies on the fertilizer
subsector, with particular attention to empirical studies. This d'ncussion helps to
understand the conceptual framework and the research problems. Section 2-2 reviews
market reform in the ON. Section 2-3 discusses the fertilizer subsidy removal. Section
2-4 discusses studies that focus on the impact of the CFAF devaluation on the
agricultural sector.
2-2- Market Reform in the Office du Niger

Market liberalization is implemented through a number of measures including
price liberalization, the promotion of the private sector, removal of quantitative and
administrative controls, reduction in the role of the state in direct buying and selling of
goods, and actions to increase the efficiency of operations which remain with the state.
The fun process, called structural adjustment, has been implemented in Mali for many
years. The adjustment policy towards liberalization and privatization started in Mali in
1981 with the cereal market restructuring program, aimed at raising farmers’ incentives
to produce more for the market (Dione, 1989). The rice market was liberalized in 1986,
and rice farmers in the ON were allowed for the first time to sell their production in the
market after deduction for water charges and credit. However, the ON was required to
defend a minimum guaranteed producer price of 70 CFAF/kg for paddy. Thus, when

the market offered less than 70 CFAF/kg, farmers sold paddy to the ON, adding to the

26
paddy the ON already collected to reimburse it for water and other inputs, thereby

creatingalargequantityofpaddytobehandled.

Dembélé (1994) reported that the ON was caught in the reform process with
thnesnucturdconsneithhemainonebeingthattheONadministrafionhadno
priormarketingexpertiseandhadtolearn howtodealwithprivaten’cetraders. The
ON was increasingly unable to fulfill the rice marketing functions as satisfactorily as
before. Cisse et al. (1993) reported that farmers in all systems were complaining about
the late removal of paddy by the ON from the villages, the late payment of the money
when paddy was removed, the nonpayment of the rebate related to the bags and the
weight differential between paddy weighed in the villages and the same paddy
reweighed by the ON at the rice processing factories. But despite all the problems raised
by farmers in the early stages of the reforms, farmers preferred to deal with the ON
instead with the private channel, which they perceived as full of risk and uncertainty.
As reported by many farmers, the uncertain open market price of rice has made the
production of vegetables (where this possible) an important enterprise in
complementing incomes (Cisse et al., 1993). However, by 1994, the situation had
changed. Diem (1994) indicated that the producer price offered by the ON (70
CFAF/kg) was not that favorable. By 1994, the average paddy prices have increased
since liberalization. It also appeared that the own-mill-sell or the buy-mill-sefl activity
was more profitable than custom milling activity. Once farmers were free to sel to
others, the small, private mills quickly outbid the ON for rice by offering prices higher

than 70 CFAF/kg. Diarra (1994) reported that, at market prices of milled rice as low as

27
CFAF 122 per kg, many private processors were still making money from own paddy

processing or buy-Inm-sell rice activity. However, major constraints were found to be
detrimental to the performance of the own milling activity. Indeed, if the quality of
paddyprocessedhappenedtobepoor,anysignificantdecreaseinthemarketpriceof
mfled rice could undermine the profitability of this activity. Storage costs combined
withtheerraticriceimportpoliciesalsoappeared to bevery immrtantfactorsaffecting
the farm prices of milled rice, the marketing costs of milled rice, and the profitability of
own paddy milling activity.

In the ON, most of the fertilizer consumed was obtained through seasonal credit
from the National Bank for Agricultural Development (BNDA) and the Dutch-financed
Arpon project, through its lending agency called “Fonds de Development Vfllageois”
(FDV). Farmers in the ON reported that the price of fertilizers supplied by the FDV was
high (5370 CFAF/50kg of urea, that was 107.4 CFAF/kg and 6160 CFAF/50kg of
ammonium phosphate, that was 123.2 CFAF/kg), whereas the price offered by the ON
for one kiogram of paddy was low (70 CFAF/kg).

Furthermore, the farmers interviewed by Cisse et al. in 1993 questioned why the
price to be paid for fertilizer was not disclosed at the time of delivery. They suggested a
decrease of the price of fertilizer if the price of paddy could not be increased. It was in
this socio-economic context that the liberalization of fertilizer market happened,
drawing more attention to the input/output price relationships, a critical determinant in
farmers’ decBion-making regarding rice production.

The complete liberalization of the input markets in 1994 stimulated the

28
participation of private dealers in the supply of fertilizer, enabling more competition and

thereby offering more choices to farmers in fertilizer supply. The problem was that in
most cases, farmers, who lack technical and functional knowledge about fertiizer,
couldn’tverflythequalityandthe righttypeoffertilher. Mafga(1993)reported that
theFDVundertakesthecontroloffertflizerquality by simplycheckingthecolorandthe
shape of the granules and by reading the documents provided by traders. However, the
FDV refuses to take responsibility for any wrong type of fertil’mer delivered to farmers.
In addition, among other constraints, inadequate financing of private firms and of
fertilizer purchases at the farm level have been identified as major problems following
market reforms (Gerner etal, 1996).

Minot (1991) addressed explicitly the impact of a fertilizer subsector reform
program in order to provide an understanding of fertilizer use in Cameroon. He
reported that privathation there had sharply reduced costs, so that farm-level fertilizer
prices rose by only 28% from 1988 to 1990. Minot reported that without any reduction
in cost, farm-level fertilizer prices would have doubled over the fast two years of the
program. However, from the point of view of the farmer, of course, this was stil a
significant increase. Moreover, with a 50% drop in coffee prices and late payment for
coffee, a substantial drop in fertilizer demand has been observed, taking the form of
fewer users and smaller quantities bought by users, implying that both the price of
coffee and the payment period were factors affecting the consumption of fertilizer.
However the computation of the value cost ratios would have shed some more light on

the issue of fertilizer demand -related problems.

29
2-3- Fertilizer Subsidy Removal

The term “ subsidy” is defined in the Longman Dictionary of contemporary
english as the money paid, especially by the government or an organization, to make
priceslowerortomakeitcheapertoproduce goods. Assuch, input subsidies remain a
sensitiveissueinthefood policydebate, not only in Mali butalsoin manypartsofthe
worfd. Gittinger (1982) defines the term subsidy as a transfer payment, that ’n, a
payment made without receiving any good or service in return (except, perhaps, the
service involved in making the transfer payment). He distinguished between two types
of subsidy: a direct subsidy and an indirect subsidy. A direct subsidy is a payment
made by a government to a producer (such as a farmer) and is a direct transfer
payment. An indirect subsidy may occur when manipulation of the market produces a
price other than that which would have been reached in a perfectly competitive market.
The benefit received by a producer or consumer as a result of this difference constitutes
an indirect transfer payment through money taken by the government from some
people to give to others. The government may deficit finance the transfer. In that case,
the resulting deficit imposes costs on others, eg., through inflation or higher interest
rates.

In Mali, some proponents of fertilizer subsidies believe that their removal
through the structural adjustment programs will pose major risks to food security (Sijm,
1992). Coulibaly (1993) reported that there are claims in the literature that the removal
of fertilizer subsidies have decreased fertilizer consumption in Mali since 1982 and

therefore contributed to the expansion of area under cultivation rather than improving

30
thepreductivityefexistingfields. Opponentstosubsidiesarguethattheiruse
encourageswastefulandmisdirecteduseofresourees becausetheprices no longerreflect
real costs (Bumb et 01., 1992). Input subsidies in this view therefore are generally
considered inefficient as a mean of increasing output and can be eliminated without any
harmful economic consequence.

Streeten (1987) reported that low prices of agricultural products are often partly
offset by subsidies to inputs such as fertilizers, credit, tractors, pesticides, seeds and the
service of infrastructure In some cases the subsidies are intended to compensate for the
protection of high-cost industries, such as fertilizer, farm chemicals and tractors. They
are then ‘d'ntortions’ that compensate for other ‘distortions’. Streeten (1987) argued
that subsidies to inputs can be useful to encourage farmers to use a new input (that h,
to help them learn what the MVP of the input is), or where external economies are
important, so that benefits accrue to others, in addition to the farmer using the input.
But subsidies also often have some undesirable effects. First, they tend to encourage the
inefficient use of subsidized inputs, for example, the waste of subsidized water. Second,
theyencouragetheuseofcertaintypesofinput, suchasfertilizer,butnotofthemost
abundant factor, labor. Consequently, they may encourage substitution of a scarcer
factor for a more abundant factor. Third, subsidies may benefit mainly larger or richer
farmers at the expense of the poor and small farmers.

The second point cannot be applied to the Malian context, where fertilizer is a
complement to labor in crop production in most parts of the country. Indeed, fertilizer

increases the yields of both crops and weeds, which leads to higher labor demand. Thus,

31
subsidizedfertih’zermay increasedemandforhhor. Thethirdpointwasobservedin

KmeaandinKenymwheresubsidiesinfactbenefitedthelargeand rich farmersand,if
accompanied by rationing, can actually deprive the small and poor farmer. Fourth, if
different crops, such as cotton, wheat and rice, use the subsid'ned input, say water, in
different proportions, the subsidy will encourage increased production of the crop using
mostoftheinput,attheexpenseoftheothers,whicthunintended result. Fifth,if
the subsidized input is exported at a profit, or smuggled abroad, the price paid by the
domestic producer can be higher than it would have been in the absence of the subsidy,
or it may cease to be available altogether.

Streeten (1987) reported that an unintended result from input subsidy policy was
observed in Bangladesh. He described the event as follows: Bangladesh reduced its
initial large fertilizer subsidy after 1979 from 10 per cent of the development budget
(and50percentofthefertilizerunit cost)to 2.4 percent(and 17 percentofthefertflizer
unit cost). This, combined with large increases in irrigation and water control
investment, resulted in more fertilizer being available to farmers, and a growth in
fertilizer sales of over 10 per cent per year, whereas before there were frequent shortages
and high unofficial prices. Renfro (1992) explained that part of the observed growth in
fertilizer consumption in Bangladesh was due to the consistently high marginal return of
fertilher to crop production and to the relatively low fertilizer price elasticity of demand.

Roth and Abbot (1990) report similar results from their study in Burkina Faso.
Roth and Abbot explained that removing input subsidies would have negligible effect on

output, demand, prices, or trade in Burkina Faso. because the marginal value of

32
fertiiserexceededtheunsubsid’uedpriceinaflregions. Gittinger(1982)expla’medthat
ifafarmerisabletopurchasefertilizeratasubsidized pficethatwfllreducehhcosts
andtherehyincreasehisnetbenefit,butthecostofthefertilizerintheuseofthe
society’s real resources remains the same. The resources needed to produce the fertilizer
(or import it from abroad) reduce the national income available to the society. Hence,
the ful cost of the fertilizer must be included for economic analyses in order to remove
the distortion from the subsidized price.

The foregoing literature shows how complex the debate is regarding fertflizer
subsidy removal on the demand and supply sides and on the level of output. How this
policy has affected the irrigated rice production system in terms of efficiency,
profitability and demand for fertilizer in rice production needs to be answered. One
important lesson we should draw from the debate regarding fertilizer subsidy removal k
that the demand for fertilizer will likely exist so long as its use by farmers remains
profitable.

2-4- The CFA Franc Devaluation

Sanders et al. (1996) argued that the devaluation in 1994 will have a short-run
effect in reducing the demand for imports of inorganic fertilizer, but a long-run effect
wfl be to increase prices of domestically produced food as imported agricultural
products become more expensive. Staatz et a1. (1994) raised the issue that “if the aim of
the devaluation is to change relative prices of tradables and non-tradables, the first
question to ask is whether it has had this effect. Have prices really changed in the

economy? If so, what has been the pattern?” A study carried out by Boughton et a1.

33
(1994)revealedthatthe50%devaluationoftheCFAfrancresultedina40%increasein

the cost of imported fertilizers, leading to a reduction of the profitability of intensive
maize. Dembélé (1996) reported that a recent studies by IFDC-Africa on restoring and
maintaining the productivity of African soils showed that “the primary effect of the
withdrawal of fertilizer subsidies and the devaluation of local currencies has been to
reduce the value cost ratios (VCR') of fertilizer use on rainfed food crops to well below
two. Heady and Dillon (1961) defined the marginal productivity of Xi, the i-th input as
foflows:
dY /dXi =bi.Y/Xi

where Y is production or output and hi the estimated coefficient with respect to the i-th
input Xi. Heady and Dillon (1961) stressed that the most reliable, and perhaps the most
useful estimate of marginal productivity is obtained by taking Xi at it geometric mean,
i.e. at the value where logXi assumes its arithmetic mean. Also Y should be the estimated

level of output when each input is held at its geometric mean. The marginal value

 

‘ The VCR is an efficiency indicator in measuring the economics of input use. For
fertilizer use we have:

VCR - MVPJMC; or VCR - FRC*( product unit price/fertilizer unit price)
where FRC is the fertilizer response coefficient. Thus, Value/cost ratio is determined by the
fertilizer response coefficient, fertilizer price paid by farmers, crop price received by the
farmers, and associated fertilizer costs (such as labor cost and credit costs) borne by the

farmers.

34
productivity of Xi can thus be obtained as follows:

By ("/05 =- biY/M Py where Ry is the price per unit of output Y.

Tessie (1996) found that recent CFAF devaluation increased the price of fertilizer
from 143 to 275 CFAF/kg, leading to a sharp decrease of VCR for all food crops in all
regions in Togo. He reported also that Togo has embarked on a Structural Adjustment
Program (SAP) since 1983. The subsides on fertilizers were removed gradually untfl
1987. This removal also contributed to the decrease of the VCRs.

In the ON, many studies treating questions related to the profitability of rice
cultivation before and after the devaluation of the local currency have been canied out,
but none of them explicitly addressed questions related to efficiency, profitability and
potential demand for fertil’ner. For example Deme (1993) provided the structure of costs
and returns per hectare of paddy production without indicating what proportion of the
costs was attributed to fertilizer. In simulating the effect of the currency devaluation,
some of the critical factors should have been taken into account. They are:

i) the frequent variation of the price of products and,

ii) the ability of farmers to buy fertilizer.

The implementation of scenarios involving the above factors could have provided
more insight into the effects of the currency change on the profitability of fertilizer in
rice production.

After the CFAF devaluation, Coulibaly et a1. (1994) suggested other exogenous

factors like rainfall and diseases were critical in explaining the level of paddy production.

35
Theyargued, forexample,thatthedecliningyieldinthe1994/95croppingseasoninthe

ONcouldnotbeattributedtofertilizersincethequantityusedperhectaredidnot
change during the indicated year, implying that the absolute price increase offertilizer
following the devaluation did not discourage farmers from using the same quantity of
fertflher as before. This implication was contrary to the conclusion of their study, that
the currency devaluation prevented farmers from increasing the quantity of fertilizer
used per hectare of rice. In figure 2, the increasing trend observed in the relative prices
of fertflirer in the ON indicated that the relative prices of fertilizer have increased from
1994 to 1997 bringing about the issues of input cost and use efficiency.

The study carried out by Del Villar et al. (1995) was intended to assess the
impact of devaluation on the income and strategies of rice farmers at the ON. One of
the findings of the study was that there was a differential impact of devaluation on the
different types of farmers in the ON. The study found that the return per hectare of rice
production has increased but was variable from one farmer to another within a given
system and from one system to another. However, which specific type of farmers won
or lost from devaluation was not stated in the report. Similarly, issues related to
effective or potential effective demand for imported goods like chemical fertilizers were
not explicitly addressed. For example, the expected increase of the level of inflation to
30% and the price of fertilizer by 15% were stated in the report as major sources of
concern about the profitability of fertilizer and other imported farm inputs. Obviously,
this concern calls for an assessment of the impacts of devaluation on the demand and

supply of these inputs.

36
Diagana et al. (1995), using a linear programming framework, reported that the
price changes brought about by the currency devaluation did not influence the cropping
pattumandtechndogychoicubypeanutandmflletgroweninSenegalDapitethh
Wing, Diapna et al (1995) suggested further analysis involving a multiperiod

programming approach taking into account the aspect of risk in the model.

37
CHAPTER III - CONCEPTUAL APPROACH OF THE DETERMINANTS OF

FERTILIZER DEMAND

3-1- Demand for Fertilizer Under Perfect-Market Conditions

Assuming that profit maxim’nation is the ultimate goal of farm business, profit
wfl be maximized where the level of inputs is set such that the marginal value product
(MVP) of variable inputs are equated with their marginal factor costs (MFC), given
compaitive markets, certainty, and no input supply constraints. In other words, profit
wil be maximum when the marginal physical product of input use is equal to the price
ratio of input and crop output (MPP=PF/PQ). The magnitude of all these variables,
hence the economics of fertilizer use, can be manipulated through fertilizer policy and
through biological research that raises the marginal physical product of fertilizer. The
difference between the MVP and the acquisition cost of the resource indicates the scope
of resource adjustment necessary to attain the economic optimum.
3-2- Demand for Fertil'mer Under Imperfect-Market Conditions

As reported by Mudahar (1978), the knowledge about yield response to applied
fertilizer forms the analytical basis for the economic analysis of fertilizer use at the farm
level, which, in turn, forms the basis for fertilizer policy formulation at the national level.
The fertilizer response is determined by a large number of factors, including crop, crop
variety, irrigation, soil quality, type of fertilizer material, management, and other agro-
climatic factors. Given the fertilizer response function, the optimum level of fertilizer use

is determined by economic factors, including constraints and risk associated with

38
fertilizer use. Given these considerations farmers will use fertil'ner only d' its use is

expected to be profitable.

Givenahudgetcomn'aintduetocashfiowpmblemsfimperfectcapitalmmket),
theleveloffertilizerusemaylowerthantheeconomicoptimum. Underthebndget
comminuthelevdoffutflizeruseunbeexpandedbyrdaxingthuecomtraino
through appropriate credit. In the short run and no budget constraint, the optimum
level of fertilizer use can be increased by lowering the input/output price ratio (eg.,
through fertilizer subsidy and/ or crop price support policies). Improved roads, by
lowering transport costs for both inputs and outputs, also raises farm-level output prices
and lowers farm-level input costs, thereby making fertilizer use more profitable.

Mudahar (1978) stressed that in the long run, fertilizer use can be increased by
raising its productivity through an upward shift in the response function (that is,
increasing MPP). This can be accomplished through developing better fertilizer
materials, better management practices, and better crop varieties with higher fertilizer
response. Finafly, water, being a key complementary input to fertilizer use, can shift the
response function through better irrigation facilities and water management.
3-3- Rule of Thumb: the VCR

McIntire (1985) argued that monetary return is the basic determinant of
fertiizerdemand. In otherwords, farmerswill notusefertilizerifit is not profitable.
Logicaly, profitability of use can be regarded as a prime factor, for farmers’ acceptance
of a particular input depends upon its profitability. Evidently, the relative level of

fertilizer price vis-a -vis agricultural product prices determines the rate of use.

39
However, the extent to which the price factor determines levels of fertilizer use and

ukimately,growthindemand,needstobeexaminedinthelightofagivensituation.
Thepriceoffertflherdoesnotaffectthedemandforitsolongasadditionalreturns
covertheadditionalcost. Intheareaswherethe rangeofprofitabilityoffertil’uernseis
sufficiently high and yet rates of fertilizer applications are low, implying that farmers are
usingfertil'neratsub-optimallevels, the price offertil'neris hardlyadisincentivetothe
farmer in extending fertilizer demand. Possibly, the level of fertilizer demand '-
significantly affected by the technological change, i.e. variations in technical coefficient
of output response to fertilizers. This study intends to examine the relevance of changes
in product and factor prices on the demand for fertilizers by farmers at the ON in the
changing environment.

Mudahar (1985) reported that the profitability of fertilizer use can be
determined by evaluating value cost ratios (VCR). The VCR is an efficiency indicator
which compares the value of the incremental crop output (MVP) due to the use of
fertilizer with the per unit cost of fertilizer used (MFC) (Tessio, 1996). It highlights that
both expected revenue and input cost determine the viability of fertilizer use (Demeke et
al.. 1995)-

Ifit is efficient to use fertilizer up to the point ofMVP = MFC, then efficiency
requires use of fertilizer until VCR = 1. However, to motivate farmers to use fertilizer in
the risky environment, some higher levels of VCR have been suggested. Tessio (1996)
reported, for example, that the minimum VCR required is 2 in order to induce farmers

to use fertilizer. For sub-Saharan Africa many authors argue that the minimum

40
acceptableVCRhastobegreaterthan2tomotivatefarmerstoadoptseed—fertilizer

technology given the level of risk involved. For Mali, for example, Sijm (1992) reported
VCRs for the fertilization of millet ranging from 1.2 to 2.1, and for rice, ranging from 2.2
to 3.8. He suggested that for poor and risk-averse peasants who have to operate under

minfedagricukuralcondifiomandhighmarkefingunceflainfiu,suchminMah,a
VCR of at least 3.0 might be more useful to encourage farmers to invest in fertiliser.

However, Mudahar (1978) argued that when the farmer makes a decision to use
fertil'ner based on MVP greater than or equal to 2 MFC decision rrrle, he or she is
making an irrational decision because the farmer can increase his profit by using more
fertilher up to a point when the MVP - MFC equality is satisfied (is, where VCR - 1).
Mudahar pointed out that farmers in Asia, Africa, and Latin America are economic
men and women and make rational decisions within their decision environment. In
their own calculations, farmers tend to use fertilizer to a point, which roughly equates
MVP with MFC, in the absence of budget and fertilizer constraints.

Mudahar provided two possible explanations for the d'ncrepancy between
farmers’ decisionruleofMVP=MFCto determinethelevel offertilizeruseandthe
perception that in determining fertilizer use farmers are guided by the decision rule of
MVP greater than or equal to 2 MFC. The first explanation is that the alleged decision
rrrle not a correct representation of farmers’ decision-making process of determining
optimum fertilizer dose. The second explanation is that the rule is correct but
oversimflifies the process of determining optimum fertilizer dose. These explanations

were tested by using nitrogen response function for rice under alternative assumptions

41
andpolicyregimeinOrissaandia)andArkansas(US). Forstandardrisk-neutral

model, the optimum fertilizer dose was determined by assuming (1) MVP-MC dechion
rule, (2) no yield or price risk, (3) no fertilizer or budget constraint, (4) no credit cost,
and (5) no other fertiizerhrelated costs. Under these assumptions, the derived optimum
ferti’nerdoseisveryhighandtheconespondingMVP/MC ratiowasone. Inorderto
test MVP>-2MC (or MVP/MC>=2) decision rule, the optimum level of nitrogen (N)
from adjusted risk-neutral model was estimated first. Next, the MVP and MC were
estimated from standard risk-neutral model by using N determined by the adjusted
risk-neutral model. The corresponding MVP/MC ratio turned out to be 1.39, which was
less than 2. Similarly, an estimate of MVP and MC from standard risk-neutral model
were made by using the optimum level of N from the adjusted risk-aversion model. The
corresponding MVP/MC ratio turned out to be 1.94 for Orissa. For Arkansas this ratio
was 1.67. In both cases MVP less than 2MC were observed. However, for Orissa the
values of MVP and 2 MC were so close that it was tempting to accept the validity of
WMC decision rule for fertilization. However, it was concluded that since the
case of Orissa could also be a mere coincidence, there was a need to estimate these
coefficients under different conditions (Crops, varieties, soil types, and environment) to
further test the hypothesis. However, Mudahar pointed out that it may not be the best
strategy to generalize the MVP>=2MC decision rule to determine farmers’ fertil'uer
adoption and use criteria. Rather all the relevant variables need to be incorporated
since their importance varies across farms, cropping systems, regions, and policy

programs. Inanycaseweneedtobearin ourmind thatthetargetVCRmayneedto

42
behighenoughtomotivatefarmersto usefertilizerin a risky environment (eg.,VCR-
2). This higher level of VCR is different from the efficient level of fertilizer use with a
VCR equated to 1.

In general in the ON, farmers face many constraints affecting their decisions to
usefertfizer. Thefirstcriticalfactoraffectingtheleveloffertilherusedforrice
productionintheONistheaccesstoseasonalcredit from formalsources(FDVand
BNDA) and from informal sources (relatives, moneylenders, traders). The second
critical factor affecting the level of fertilizer used involves various factors like the price of
the fertilizer itself, the price of paddy in the market, the availability of family and hhed
sources of labor, the size of the rice plot, the quality of the rice plot, the level of the
interest rate, and farmers’ perception about risk.

Another serious constraint facing the fertilizer delivery operations in the ON
zone is the poor condition of the rural roads during the rainy season. This makes
difficult, if not impossible, the delivery of fertilizer to farmers at the right time, right
place and right quantity. In most cases fertilizers are delivered many months ahead of
therighttimebecauseoftheinaccessibilitytofarmingareasattherainyseason. Then
fertiizers are stored in poor conditions before their application. During the storage
period, the quality of the fertilizers decreases while at the same time the interest
payment increases, aggravating the high default rate in credit reimbursement. For
example, Weijenborg (1993) reported that the loan recovery rates (percentage of loans
due) by the FDV agents in the ON were only 1% and 5% in February and March 1993,

respectively. As a consequence, the amount of overdue loans in 1993 cropping season

43
accountedfor44% ofthetotalamountofcreditdisbursed, reducing drasticallythetotal

liquidity for lending. Weijenborg also reported that in 1992, the rate ofrepayment of
theloand'nbursedbytheBNDAintheONwassolowthattheBNDAexpresseda
serious concern about its lending activity in the future. Both the FDV and the BNDA
were seeking the best strategy to overcome the problem of low repayment rates and
cutting down theh' operating costs. It becomes clear that any improvement in the
logistics and transportation systems and credit arrangements may help improve
coordination and productivity.

Another factor, which can be a potential constraint to the demand for fertilizer is
that most of the rice production systems in the ON are not fully, restored. This creates a
high risk related to fertilizer use and its payoff. Moreover, the low rate of credit
reimbursement by most farmers in the ON prevents them from gaining access to
fertilizer from the FDV.
3-4- Risk Considerations

The amount of fertilizer that any farmer will use will depend on anticipated yield
response, expected product prices, fertilizer costs, capital stock and/or credit availability,
the degree of risk and uncertainty that the farmer must take and his ability to absorb
such risk (Falusi, 1973).

Given the input prices, returns from fertilizer use vary greedy on account of
changes in size of response due to variations in climatic and physic al factors. Physical
properties of soil, availability of supplementary nutrients, and climatic and rainfafl

conditions affect the size of response to a considerable extent. Among these, a close

44
rehfiomhipexhobetweenmmfaflandcropresponutofertflizeruseevenintheON

during the pre-irrigation period.

Mudahar (1985) reported that the factors contributing to low fertfl'ner demand
include (1) low fertilizer response, (2) high fertilizer cost, (3) low crop prices, (4) high risk
oflosing money as a result ofthe variability in fertilizer responseand prices, (5) lackof
cash or credit, (5) lack of knowledge, and (7) lack of complementary farm inputs such as
fertilizehresponsive crop varieties, water, and insecticides. Gerner et al. (1996) found in
Africa that in addition to aggregate fertilizer supply constraints, high procurement and
distribution cost, timely availability of the right types of fertilizers, foreign exchange
constraints, lack of adequate financing of private firms and of fertilizer purchases at the
farm-level have been identified as major problems following market reforms. In
Ethiopia, Demeke et al. (1996) found that there is a relationship between input market
and grain market performance. In other words, the performance of the grain marketing
system in Ethiopia strongly influences the profitability of fertilizer use by farmers. They
concluded that efforts to reduce grain-marketing costs should be viewed as a critical
component hr the overall strategy to stimulate fertilizer demand and crop productivity.

In the regions where irrigation facilities are inadequate, for example in the semi-
intensive and non-restored area in the ON, the amount and distribution of rainfal
influences the level of fertilizer use considerably. Moreover, where fertil'ner application
is spread over two to three of applications, ie. as a basal dose and top dressings, the
amount of rainfall has a greater impact on rates of applications. In general, use of

fertil'ners varies with amount and regularity of rainfall.

45

Intheconditionsofirrigatedfanning,forexampleintheintensivesystem ‘mthe

ON, where availability of water can be regulated as per cropping schedule, yield
uncertainty h considerably reduced. Secondly, intensity of cropping increases under
irrigated conditions; and fertilizer needs are felt more on account of continued cropping-
sequence or multiple cropping. Thus, the response to fertilizer use is higher when
irrigation h available. Therefore, the extent of irrigation plays a vital role in
determining levels of fertilizer use. Rates of application of fertilizers increase with
cultivation of higher yielding varieties, and the extent of adoption of these varieties
stimulates growth in fertil'aer demand.
The factors affecting the use of fertilizer at the ON have been discussed in section 3-
3These factors are sources of high risk and uncertainty related to the use of fertilizer and
its payoff. One of the consequences of this problem is that the recommended levels of
fertilizer from agronomic experiments are not applied at the farm level. Most farmers in
the ON were feeling insecure after the removal of the guaranteed marketing outlet at
official producer prices. Indeed, prices of input and output fluctuate with the changes
in supply and demand. Rice yields vary substantially with the variable level of rainfall
and irrigation water, the outbreak of weeds, insects and diseases.

The occurrence of the foregoing events results in income variability. However,
one of the challenging objectives of any household head in any year in the ON is to
obtain a minimum income from rice production to meet the household’s financial
obligations. Another challenging objective for any household head in the ON in any

year is to meet the minimum level of rice self-sufficiency. The quantity of rice needed to

46

cover the minimum level of self-sufficiency varies across systems in the ON and can
reach on average 40 per cent ofthe total rice production (Cisse et al., 1993). This
consideration must aho be included in the analysis.

3-5- Financial and Economic Analyses

F'mancial and economic analyses are complementary in that the financial
analyst takes the viewpoint of the individual entrepreneur and the economic analysis
that of the society as a whole. Despite the complementarity between the two concepts,
three important distinctions must be kept in mind (Gittinger, 1982).

First, in economic analysis taxes and subsidies are treated as transfer payments
to or from the government, which acts on behalf of the society as the whole, and are not
treatedascosts. Inf'mancialanalysistaxesaretreatedasacostandsubsidiesasa
return.

Second, in financial analysis, market prices are normally used. These take into
account taxes and subsidies. From these prices come the data used in the economic
analysis. In economic analysis, however, some market prices may be changed so that
they more accurately reflect social or economic values. These adjusted prices are called
“shadow” or “accounting” prices.

Third, in economic analysis, interest on capital is never separated and deducted
fromthegrossreturnbecauseitispartofthetotal return tothecapitalofthesocietyas
a whole and because it B that total return, including interest, that economic analysis is
designed to estimate. In financial analysis, interest paid to external suppliers of money is

deducted to derive the benefit available to the owner of capital. But, interest imputed or

47

“paid” to the entity from whose point of view the financial analysis is being done is not
treatedasacostbecauetheinterestispartofthetotalreturntotheequitycapital
contributedbytheentity. Henceithapartofthefinancialreturnthatentityreceives.

Barry (1994) reported that in most developing countries, it is common to note
that mources are not allocated efficiently because either input markets or output
markets, or both, function imperfectly, owing to not only market failures, but also to
government interventions, through its fiscal and pricing policies. Examples of
government interventions are protective tarifi's, import bans, pan-territorial and pan—
seasonal prices. With such interventions, market prices may differ from social
opportunity costs and government-induced prices may lead to suboptimal resource
allocation. In this respect, private profitability may differ from social profitability,
which is the true measure of the efficiency of resource allocation because inputs and
output are valued at their opportunity costs or shadow prices.

Assuming that the domestic market prices of agricultural inputs and products
were distorted before the CFAF devaluation in 1994 means that these prices did not
reflect their scarcity value (social opportunity cost) because of government intervention.
The economic analysis uses prices from which all market distortions or taxes have been
removed. Ah subsidies and taxes are considered as transfer payments between groups of
producers or consumers in the same country. Furthermore, if any inputs or products are
imported, then not only must taxes and subsidies be removed in the valuation of these
goods but also an adjustment be made for the rate of exchange.

Stryker er al. (1987) found that the Malian local currency was overvalued by 33-

48
37 percent during 1981-1985 period. The method employed consisted of correcting the
leveloftheofficialeachangeratewo)byaddingatermthatadjustsittotheratethat
would need to prevail in the market for foreign exchange ifthere were no current
account deficit.

For small deviations, this equilibrium rate of exchange E' can be approximated

by:E' - r. + 1:. (DEF/(e.EXP + am»

Where L is the offic'nl exchange rate, DEF is the current account deficit, IMP is the
exiting level of unports, EXP is the existing level of exports, e. is the elasticity of supply
of foreign exchange, and e. is the elasticity of demand for foreign exchange. The values
of these elasticities were roughly estimated as e. = 1.0 and e. =- 2.0. Stryker at al. (1987)
argued that if the distortions in domestic prices resulting from price controls and from
trade taxes and controls were accounted for, the exchange rate would almost certainly be
even more overvalued because of the high tariffs and the system of import controls.

Salinger and Stryker (1991) argued that in Mali, like in all CFAF countries, for several
years, the equilibrium exchange rate has been above the observed or official exchange
rate because of unsustainable current account deficits and trade policy distortions which
resulted in an excess demand of foreign exchange and led to extra borrowing and
excessive drawing down of foreign exchange reserves. The corresponding overvalued
exchange rate made imports, such as agricultural inputs, cheaper (ie, less domestic
currency paid out for imports) and the price of exports as well as the domestic prices of
non-tradables, particularly labor, more expensive (ie, less domestic currency earned by

exports). In other words , Malians have paid a premium on traded goods over what

49
they paid for non-traded. A rate of about 50 percent overvaluation of the CFAF has

been reported by Salinger and Stryker (1991) whenever the deficit was more important,
accounting for about 44 percent of the overvaluation. Thus, because the CFAF
currencywasovervaluedbeforethe l994devaluation,itwouldbeneceuarytousean
adjustedratetoconvcrtthepriceofgoodstradedinforeigncurrencyintoadomestic
equivalent in undertaking an economic analysis for years prior to 1994.

Economic analysis corrects the distortion in order to undertake any evaluation
on the basis of the opportunity cost to the country as a whole of the resources invested in
me activities. For non-tradable goods such as labor, their social value is found by
estimating their social opportunity cost (ie, the net income foregone because the factor
is not employed in its best alternative use). In contrast, for tradable goods such as
fertilizers or paddy, the appropriate social value should be based on world prices
(expressed in domestic currency) because these prices represent the society’s choice to

permit consumers and producers to either import or produce those goods domestically.

50
CHAPTER 4 - EMPIRICAL MODEL

44- Introduction
Thischapterpresentstheanalyticalmethodthatwillbeusedinthepmpoeed
study to the effective demand of fertilizer in the ON areas. The chapter consists of six
sections, and is organ'ned as follows. The introductory section is followed by a section
that is devoted to a critical review of methods employed by researchers for estimation of
consumption of fertilizers. This section provides a justification for the selection of the
mathematical programming approach. The third subsection provides a review of the
conventional mathematical programming approach followed by section four, which
covers the parametric programming technique. Section five discusses the models to be
implemented in the study area. This section includes the expected results, the method of
aggregation of the potential demand for fertilizer, and the method of calculation of the
import parity prices for fertilizer. Chapter 4 ends with a section describing the
modification of the basic model to incorporate risk.
4-2- Review of Methods Employed for Estimation of Consumption of Fertilizers
Maharaja (1975) grouped the procedures researchers have used to estimate

fertilizer consumption into three types:

5 1
(l) Need-based approach

(2) Area-crop-coverage approach; and

(3) Regression approach.

Theessenceoftheneed-basedapproachistoplanforneededexpansionin
agricultural production. Fertilizer requirements are calculated by using specific input-
output ratios in order to achieve the required or desired quantity of additional
production.

The essence of the area-crop coverage approach is to determine fertil'uer
requirements by considering probable trends in cropping patterns and expected
coverage of area under fertilizers at recommended doses.

The broad framework of the regression approach is the prediction of fertil'ner
demand based on time- series data and by accounting for the influence of one or more
variables on fertilizer consumption. In the regression approach, the estimation method
varies fiom trend fitting to multiple regression analysis based on econometric models.

The need-based approach virtually ignores the profiles of demand viewed from
the farmer’s angle. Such estimates are based on a priori assumptions; for neither are
they based on considerations of agro-climatic influences on growth of demand nor are
they assessed in the context of factors influencing farmers’ ability and willingness to use
fertilizers at different levels and in different proportions. It is always possible to deduce
from simple arithmetical exercises the quantity of fertilizers that should be used at

specified average levels to achieve certain levels of output. However, such estimates

52
ignerethebasiceconomicrealitiesatthefarmlevel.

The area—crop approach assumes rates of application of fertilizers at near-
optimum doses or at recommended doses. The estimates arrived at by this approach
overiookcurreuttrendsintheratesofapplicationoffertilizersondifi’erentcropsandthe
rateshthefuturethatarelikelytobeadopted duetovariousdynamicfactoraThe
morsfromthismethodmfleninthednecfionofovmestimafiomcanbecosfly.

Econometric models identifying factors affecting fertilizer demand and
quantifying their significance for estimating demand at macro-level could provide more
realistic picture. Yet these econometric models can be inadequate on account of
collinearity problems among explanatory variables, causing the impossibility of
separating the effects of one component from another. The results from econometric
analysis may be sensitive to practical data availability and variable construction
problems. Indeed the analysis requires adequate timeseries or cross-sectional data for
al the relevant variables. Statistical problems aside, econometric analysh of historical
data to predict future responses is not without its difficulties. For example, in predicting
the outcome of a specific promm proposal, econometric models are likely to he
inadequateasapredictive device ifthe new program provides anewsetofinstitutional
restraints for, which there is no historical counterpart to use in estimating response.
Staatz (1997, personal communication) reported that one of the biggest limitations of
the econometric approach in the setting of countries like Mali where there have been
hrge changes in the rules governing the markets is that parameters estimated using

historical time series may be very poor predictors of future behavior in the new market

53
structure. A mathematical programming approach can overcome this problem by

My simulating the current situation.

Indeed a mathematical programming based on the macro-economic and micro-
economic environments and followed by a sensitivity analysis or parametric
programming can produce better results. The programming approach has the
advantage of being able to derive optimal production and consumption plans, etc. to
satisfy a multiplicity of objectives, of taking into account the role of price, and of
specifying in considerable detail the constraints under which production, income
generation, and policy making are taking place (Sadoulet and De Janvry, 1995). In
recent years, programming models have been used extensively to address many types of
policy questions, including input demand analysis. The basic approach has been to
validate the model for a base period, and then use it to simulate adjustments and
responses of economic agents to policy changes (McCarl and Spreen, 1980).

In deriving demand estimates by parametric programming procedures, a series
of related problems are solved in which the price of the factor of interest is varied from a
min'uuum to a maximum level. As the price of the factor varies, different levels of input
use become optimal and, in consequence, a series of price-quantity relationships is
developed. Flinn (1969) argued that the synthetic demand function derived by the
price-quantity relationship procedure is of a stepped nature because of the linear nature
of the production data, and the finite number of production alternatives and resource
restrictions considered. Therefore, the demand ‘curve’ derived by using linear

programming differs from the smooth curves of conventional theory. Flinn (1969)

54
condudedthattheduiveddemandcuwaatbestcanonlyberegardedushoumn

estimates due to uncertainty about future prices, technologies and institutional
constraints which may be imposed on the system. Even in the short run, farm
managers’ decisions may vary substantially from the actions predicted by the linear
programming models of the farrrr firm. In particular, different subjective estimates of
managers in relation to crop yield and prices, and different attitudes may result in
farmers’ actual decisions differing, somewhat markedly, from those indicated as
optimal. In this study, the basic model will be modified to incorporate the risk
component. This procedure is discussed in subsection 4-5-5. The next section provides
more detaik about the conventional programming model, followed by section 4-4, which
d'ucusses sensitivity analysis or parametric programming.
4-3- Mathematical Programming

Rae (1977) described mathematical programming as a planning tool used to
determine the best plan or course of action among which:

(a) there are many alternatives for the plan;

(b) a specific or numerical objective exists;

(c) the means or resources available for obtaining the objective are
limited.

The strength of mathematical programming lies in its ability to handle a large
number of interrelated variables and thus to cope with peasant farming systems that
are characterised by a high degree of interdependence between production and
consumption, consumption and investment, investment and resource availability, and

social and cultural constraints (Low, 1974). Beneke and Winterboer (1973) steered

55
that the great advantage of programming is that it allows one to test a wide range of

alternative adjustment and to analyze their consequences thoroughly with a small input
of managerial time. The question “what would happen if...?” can be posed repeatedly
andanswered rigorouslyand quicklyoncethemodelisbuilt. In atypicalprogramming
analysis, the magnitrrdes of the marginal value productivities of fixed resources are
obtained as by-product of the conventional programming solution. The element in the
LC row of the disposal activity columns represent the marginal value products of these
resources and are regarded as the measure of the ceiling that should be set in acquiring

extra resources.

The conventional programming model can be formalized as follows:

MaxZ-C’X (1)
subject to AX s B (2)
and x 2 o (3)

where: Z = the value to be maximized

C = n by 1 vector of prices

X - n by 1 vector ofactivity levels

A - m by n matrix of input-output coefficient

B =- m by 1 vector of variable factors or other restrictions.

In a programming model, activities can be grouped into (at least) five categories.

Theseare:

56
(1) production activities;
(2) buying activities,
(3) selling activities;
(4) storage activities; and (5)
(5) transfer activities: these activities provide a vehicle whereby the service

or output of one activity may be transferred in the model to another

activity.

The number of activities depends on the availability of data and on the objective
of the study. It is important to note that large and complex models are costly to develop
h terms of both time and money; and it is not always certain that the benefit to be
derived from using a more sophisticated model (in terms of greater precision of the
planning decisions derived from it) are sufficient to justify the cost Furthermore, the
solution of the programming analysis rests on the validity of the following assumptions
(Barnard and Nix, 1973):

1) all the enterprises under consideration must be linearly additive, excluding
the possibility of interaction in the amount of resources used per unit of
output, whether or not enterprises are produced alone or in various
proportions.

2) resources used (such as land, labor and capital) and the commodities
produced are infinitely divisible.

3) a limit exists to the number of alternative enterprises and resources, which

need to be considered.

57

4) it is usually assumed that resource supplies, input-output coefficient, and
prices are known with certainty. This assumption of single valued
expectations may seem unrealistic for the farming situations in Mali.

5) the large element of fixed resources usually incorporated in the linear
programming matrix means that relatively short-term situations are being
studied. When solutions are not widely different from the current farm
organization, they may represent the farmer’s problems adequately. When,
however, they differ widely, several years may be required to attain the
suggested organization, as is true of longer-term planning in general. In such
circumstances, it may be preferable to use a technique in which a dynamic

element is incorporated.

Despite the limitations of the programming technique, it advantages outweigh
the limitations. Examples of the use of mathematical programming for farm planning
are many, including applications to planning peasant farms (eg., Clayton, 1964; Beyer,
1971; Traore, 1979; Niang, 1980; Ogunbile, 1980; Etuk, 1982; Kamuanga, 1982; Mafga,
1983, Cisse, 1987, Camara, 1988, Ngwira, 1994).

Bernard and Nix (1973) outlined the main techniques developed to deal with some of the
conceptual criticisms of linear programming. These techniques and the inadequacies of

‘ordinary’ linear programming which are tackled by each of them are presented below.

58

Table 3: Other Computer Programming Techniques

 

 

Techniques Shortcoming of Linear programming tackled by each
technique

Inteser Pros-immins

Separable Programming Structural

Monte Carlo Method Structural; Objectives

Parametric Programming Structural and Uncertainty - but only to a very limited
extent

Dynamic Dynamics; Objectives - to a limited extent

Linear Programming

Dynamic Programming Dynamic Primarily; Structural and Uncertainty also

Quadratic Programming Objective and Uncertainty

Stochastic Programming Objectives and Uncertainty

Game theory Objective and Uncertainty

 

Source: Adqptedfi'om Barnard and Nix (1973)

On the table above, structural refers to the assumptions of linearity and

continuity in a linear programming model. The latter means that ‘integer’, or

indivisible unit, cannot be dealt with in a fully acceptable manner. In addition, a linear

programming optimal plan may not suit the longer-term aims of the farmer. Al these

59
problem of longerme planning require a ‘dynamic’ treatment rather than the
‘static’approachoflinearprogramming. Asdiscussedinpoint4above,linear
programming’n‘deterministic’,thatis,itassumessinglcvaluee fornetrevenues,
constaintandresourcerequhement,asthoughperfectknowledgeexisted. Sincean
hpertant aspect of a farmer’s objectives relates to his attitude to risk and uncertainty,
BanardandN'u(l973)stessedthatfinearpmgnmmingonlymaximhuwithinthe
limits imposed and with the data given; thus allowances can be made for personal
preferenceeand,tosomeextent,forrisk

Kamuanga (1982) used a one-period linear programming model to evaluate the
profitability from the introduction of five improved practices or intensification
techniques in the ON in 1982 under the current paddy price level of 60 Malian
W. The result ofthe study indicated that across all LP runs, there was a strong
indication that the ON should concentrate the intensification program on smal and
medium sized farms. Since then, the ON has been subject to many changes which have
been described in chapter 1]. Therefore, it becomes necessary to re-evaluate the
profitability of rice production and to compare and contrast the outcomes with previous
result. It is expected that such analysis will provide farmers, researchers, and policy
makers with more insight into the resource use efficiency, profitability and fertilizer

demand tsues after the market reforms.

 

’ZMF-ICFAF

4-4- Parametric Programming

Parametric programming is a modification of the conventional model to allow
the implementation of variable price programming and variable resource programming.
It is a technique that aflows a series of optimum plans to be estimated, for differing
leveh of any parameter of the problem (Rae, 1977). Such parameters may be product
prices, variable factor costs, crop yields, supplies of fixed factors, or the requirement per
unit of any crop for any fixed factor. 'In short, parametric programming enaqu us to
know how optimum farm plans change as prices, input/output coefficient, or resource
endowment vary.

Ogunfowora (1972) has conceptualized a programming problem with a

parametric objective function as follows:

max Z, = chXj
j=l

where: Z Z (X1, Xwaxjs X.)

L: = the objective function
to be maximized for a given price level within the
acceptable price range;

b.‘ = theleveloftheith

resource available.

Assuming that:

61
l) C,’ and C,” = the lower and upper

limits of the price of the jtls activity and Cj’s C, S C,”;

2) O - constant increment in
the price of the jtlr activity;
3) K = the number of

optimum solutions within the price range;

We can write: (C1’°Cj”)/9 ‘ k

Cf- Cj’ ‘ 9K

This approach is useful because it enables a model builder to determine the
number of optimal solutions and the levels of increment in price of the j... activity within
an acceptable framework.

For variable price programming, the objective function is parameter’ned with
respect to the price of the activity of interest. Optimum plans are then derived for each
price level. The variable resource programming is analogous to the variable price
programming. In this case it is the resource levels that are parameterized.

Parametric programming can be used to derive product supply and factor
demand functions (Rae, 1977). For a given type of farm input, the farmer’s demand
function for this input can be derived by plotting the quantities that should be

purchased against the various levels of prices. The stepped appearance of the graph

62
ehtainedfiomthtpmcedunisaresuhoftheuuofahneaflysegmentedproducfion—

pouibfityboundarybythefinarpmgrammingmethomandcanheconsidaeduan
approximation to the smooth demand functions of economic theory (Rae, 1977). The
basic principle is the profit maximization principle introduced in section 3-1: if a farmer
coneidasX.toheavariableinputandwishestodeterminewhetheritworddhe
profitableforhimtoaddto,orreduce,hispresentuseofthisfactor,hewflneedto
compare the marginal value product (MVP.) given his present supply of X. , with the
price per unit of this factor, P(X.). Only if MVP. exceeds P(X.) will further purchases of
this input be profitable, and the most profitable level of X. wil be that for which MVP.
equals P(X.). Thus the graph ofMVP. against the use ofX.'may be interpreted as a
farmer’s demand function for this input since it indicates the quantity that he should
purchase at various factor prices.

By programming a number of ‘representative’ holdings in an irrigation area, for
example, demand functions for irrigation water can be derived for each holding (Flinn,
1969). Given knowledge of the supply situation with respect to irrigation water, a
pricing policy can be formulated so that the total regional demand for this resource can
be equated with the regional supply. Alternatively, by programming a number of
representative holdings, supply curves could be generated by varying the price of the
product of interest. We could then determine total regional output of that product at
various price. This information would help in the making of pricing decisions ( Martin,
1988).

Kottke (1967) concluded from his examination of the anatomy of a step supply

63
function that it is a valid approximation of agricultural supply behavior, particularly at

the firm level. However, in terms of demand, Yaron (1967) from his empirical analyst
of the demand for water by Israeli agriculture, found that the shape of the demand
function for water was highly dependent on the sociopolitical economic mix, which
determined the framework within which it is derived. Therefore, Yaron deduced that
the more comprehensive and realistic was the agricultural development program or
projection available as a background for the derivation of the agricultural demand
function for water, the more realistic its estimate. The implication of this finding is that
the derived demand and supply curves should be interpreted with a minimum dose of
good sense. The results from these types of analysis are highly specific to the area under
study. These limitations apply to the empirical model to be implemented in this
proposed study and the result, which may be derived.
4-5- Empirical Model, Expected Results, and Modification of the Basic Model
4-5-1- Empirical Model for the Study Area

First, an empirical model based on the conventional programming framework
wfl be built to represent a typical household in each rice production system. The

structure of the farm model is as follows:

The objective function is maximized subject to the following constraints:

0 five land constraints,
0 nine monthly labor constraints from May to January,

0 ten constraints on the average yield permitted,

0 one minimum subsistence and income requirement constraint,

one fertilizer supply constraint. Fertilizers include urea and ammonium
phosphate and are considered separately in the model.

0 one seed supply constraint related to rice seeds used on ON plots.

0 one organic manure supply constraint: although organic fertilher
requires no outlay of cash, collecting, transporting and applying organic
fertilizer are extremely labor-intensive. Organic manure includes human and
animal manure, decomposed grasses and rice straws and household waste
product.

0 one animal feed supply constraint,

0 nine capital constraints and /or transfer rows.

Activities:
Fourteen groups of activities are defined: rice production on ON and outide ON

plots; horticulture (onion, tomato, garlic, pepper), root crop production (sweet
potatoes); cereals production (millet and sorghum in mixture, maize as sole crop); feed
buying activity; labor hiring activity (labor hired in the household, labor hired out of the
household); exchange labor (through village level associations and among individual
farmers for land preparation, weeding and tillage; off-farm employment (in the village
and/or in the city); equipment hiring (oxen traction team for a work day for land
preparation); capital transfer from May to January to meet farm expenditures; selling

and buying activities of rice, vegetable crops, cereals; consumption activity of rice

65
(cultivated outide ON plot), cereals, and sweet potatoes; input supply of rice seeds,
fertiiners (urea and ammonium phosphate), organic manure and feeds for draft cattle;
borrowing activities from BNDA, FDV and informal sources. None of the activities
discussed can be operated at negative levels.

Rice‘vegetable crops and cereals arethe majorcrops grown intheON andwl
beconsideredasthemainproductionactivities,withaunitofonehectare(ha)inthe
model. Double cropping of rice will not be included in the model because the period
considered for the proposed study start from May and ends in January, which
corresponds to the beginning of this off-season activity and the end of the proposed
survey or study.

Like rice, vegetable crops are cultivated in every zone in the ON. Table 4 gives
an idea about the relative importance of vegetable crops (including sweet potatoes) in
terms the percentage of households involved in this activity.

Table 4: Relative Importance of Vegetable Crop Production (Percent of households)

 

 

 

 

 

 

Niono Sahel Molodo Kokry
Vegetable
ARPON Non-restored Non-restored ARPON
Area Area

Onion 51% 99% 59%
Sweet Potato 34% 20% 13% 16%
Pepper 25% 7% 33% 16%
Tomato 18% 13% 25% 24%
Garlic 47% 13% 5%

 

 

 

 

 

 

 

Source: Adapted from IER’S 1989/90 Survey.

66
Foranyfertiizerprocurementactivity,thefarmgateprice perkilogramandthe

quanfityofthatfatflizeruappfledbyfamasperhectamancomidaedinthemodd.
Inotherwords,the quantity ofanytypeoffertilizeris the average quantity ofthattype
offertilizerusedtocultivateonehectareofriceon ON plot. Itisassumedinthemodel
thatthefarmercanbonowmoneyfrom BNDA,FDVandfromaninformalsouree
(moneylenders and friends) at an interest rate (in cash or in kind) to buy fertilizers and
other supplies, to hire labor at the going wage rate, and to hire equipment when the
need arises at any time from May to January. Table 5 below gives an idea about how
the ferti'ner buying activity will be accounted for in the LP matrix.

Table 5: Incorporation of Fertilizer Buying Activity

 

 

 

 

 

 

 

 

 

 

 

Activities Rice Buying Buying Ammonium Signs
Resources (ha) Urea Phosphate
(R8) (R8)
Fertilizer Buying - -
Urea + -1 Equal 0
Ammonium Phosphate + -1 Equal 0
Source: From flee author

NB: The pins (+) signs on the table correspond to positive required levels of fertilizer per
hectare of rice. The minus (-) signs are the unit prices of fertilizer. The fertih’zer

acquired is transferred to the equality constraints via the -1 signs.

It is also assumed in the model that money earned from off-farm emfloyment

may have a positive impact on the use of fertilizers and the level of rice production on

67
ON plot, via provision of liquidity to buy inputs. In other words, off-farm employment
providescashtothehousehold,whichmayheusedtopurchasefutilherifituseh
profitable.

For rice and vegetable crop output marketing, the selling activity in the model
convert physical output into cash via sale. The consumption activity transfers rice, root
crops and cereals output from the farm storage to meet minimum consumption
requirement. The consumption activity is built into the program to make sure that rice,
other cereals, vegetables and sweet potatoes consumption habit are respected and that

the subsistence requirement in term of calories are met. The activity unit is 1 kilogram.

Resources Available:

Land: The model includes five types of land that are:

1) Rice lands inside ON, where improved varieties of rice are the only crops
grown with irrigation water. This type of land is commonly known as
“Casiers” and is under the supervision of the ON authority.

2) Rice lands outide ON, where only traditional varieties of rice are cultivated.

This type of land, called “Hors easier”, is not under the supervision of the
ON authority. However, the “Hors easier” can also enjoy irrigation water
from the ON as long as the water fees are paid by farmers. No fertilizer is
applied on these lands because of the cultivation of traditional varieties of

rice and the poor degree of water control.

68
3) Lands for vegetable crops production: Horticulture is practiced on this type

of land, which is outside the rice plot and can enjoy irrigation water subject
to the payment of water fees.

4) Lands for coarse grain (millet, sorghum, maize) production: This type of
land is located some distance from rice and vegetable crops production areas.
Lands for coarse grains production don’t enjoy irrigation water from the
ON. Crops produced on these lands are totally rainfed.

5) Lands for root crops production: This type of land is similar to land for

coarse grain production.

Labor: The model includes four types of labor including family labor, hired labor,
exchange labor, and off-farm employment. The total farm workdays of eight hours per
day available on an average household in each system will be calculated by converting
al categories of labor in each month to a person-day equivalent. Norman’s weighting
formula will be used to convert family workers into adult man-day equivalent. Smal
child (under ten years old), large child (above ten years old), female adult, will be
converted to adult man-day equivalent (before totaling ) by using Norman (1973)
weighting formula of 0.00, 0.50, 0.75 respectively. In Mali, it has been noticed above 60
years, some people stil do some very useful work. Hence the weighting rate of 0.50 wil
be used to convert this category of labor force(before aggregating) to adult man-day
equivalent. Hired labor, exchange labor and off-farm employment will not be counted

in the household labor available.

69
Consumption Requirement:

Farmers in the ON grow traditional varieties of rice, coarse grains and root crops
essentially for their own subsistence requirement. Improved varieties of rice are grown
forsaleonON plot. Vegetablecropsareessentiallygrownformarket,althoughpartof
them h consumed to meet household subsistence needs. A typical household in the ON
can be considered a production unit for profit maximization on one side and a
consumption unit on the other. Elsewhere in Africa, for example in Nigeria, D.W.
Norman (1973) has shown in a study of small farmers that profit maximization and
food security were not in conflict. While the provision of adequate food for the family
was given top priority, it was found that the pattern of resource allocation was consistent
with profit maximization. Martin (1988) assumed in Senegal that besides the profit
motive the other important component of the farmer’s objective function is the food
secruity objective. This objective pushes him/her to grow food crops for home
consumption and to select crops for sale in order to guarantee a minimum income
whatever the state of nature. Both actions may run counter to the profit maximization
objective. Therefore, the farmer often has to make trade-offs among conflicting
objectives. Thus Martin (1988) imposed several constraint on his model to reflect the
priority given to food security by Senegalese’s farmers. In the ON case in particular, the
heads of households have the social responsibility of ensuring that the food needs of al
members of the household are satisfied. Therefore, an attempt to introduce realism into
the model will be made by maximizing the objective function within the framework of

consumption patterns. In other words, besides the selling activities, consumption

70
activities w'fl be built into the model to make sure that subsistence crops consumption

habit are met. That t, subsistence crops are consumed and are expected to fulffl the
minimum level of calorie requirement for individuals in the typical household in each
system. TheaveragesiseofhouseholdintheONareasis 10persons (Cheetah, 1993).
An average household in the ON must consume 2124 kg ofcereals peryear to meet the
minimum requirement of 2300 calories per person per day. Further discussion of food

security issue in the ON is provided in section 4-5-5.

Operating Capital:
Operating capital includes all production expenses on fertilizers, hired labor and
cost for draft cattle maintenance and rice seed cost. It is assumed that farmers in all

systems have access to the same lending sources.

ffii

The input-output coefficient (aij’s) express the amount of input i needed for one
unit of activity j. For land, the coefficient is one for all corresponding enterprises. For
labor, the average coefficients per hectare will be used. In other words, the ratio
obtained by dividing the total number of man-days spent on an enterprise within a
particular month by the total area of land in hectares allocated to the enterprise in
question, is the input-output coefficient.

Production coefficient will be derived by dividing the total output in kilograms
by the total area in hectares planted with the specific crop or crops combination like

millet and sorghum in mixture. Operating capital includes expenditures on fertilizers,

71
hiredlabor,draftcattlemaintenanceandriceseedcost. Fertilizersusedperhectareof

ricefimmthepficcpakflogramofeachtypeoffenflherwiflbeenteredinthemodd.
Thecedofhiredlaborwillbecapturedineachmonthatthegoingwagerate.
Expenditmuonriceseedsandfeedswflbeenteredinthemoddunderthemonths
durhrg which they occur.

4-5-2- Expected Results

The basic model will be run using two types of price vectors: price of rice and
fertilizer after the CFAF currency devaluation in January 1994. The aggregate demand
for ferti’luer that needs to be satisfied by the private channel for input delivery wil be
calculated follow’mg a method described in subsection 4-5-3.

In the analysis, it will be assumed that the 1994 CFAF devaluation removed all
the distortions from this currency to reflect its social value. Therefore, no adjustment
wil be needed to realigned the official exchange rate prevailing in Mali afier the
devaluation.

It is expected from the analysis that we will get an estimate of the aggregate
demand for rice production that needs to be satisfied by the private channel for input
delivery.

In order to assess if the ON can support a private fertilizers distribution system,
the leveh of aggregate potential demand for fertilizers will be estimated from the model
by using the import parity price of fertilizer at different places at the ON.

4-5-3- Aggregate Potential Demand

Classically the demand schedule for a given input in a competitive market h

72
derived through horhontal summation of the demand schedules of the individuah

producers in the region being studied (Flinn, 1969). Aggregation can be performed in
this manner providing two conditions are met. East, the various producers in the region
must confront the same factor prices. Second, only the price of the input of interest is
varied; alotherpricesareassumcdto remain constant.

The evaluation of the aggregate potential demand for fertilizer in the ON is based
on the optimal levels of fertilizer and the total number of farms. The that step is to
calculate the consumption of fertilizer by multiplying the optimal fertilizer levels using
different input-output price vectors by the number of farms in each of three rice
production systems. The second step is to evaluate total fertilizer consumption at
different price levels in the ON by adding consumption from all the production systems.
More explicitly, the total demand schedule for fertilizers at the ON can be derived as the
aggregate of the demand schedules for the representative households defining the

population. The aggregate demand schedule in the ON can be specifred as:

73

where: D is the aggregate demand for the ON area;
ni isthenumberofhouseholdsintheithstratum;
fi istheoptimal fertilizer input fortheithrepreseutativehousehold
when the price of fertilizer is pf; and

t is the total number of strata specified for the ON.

The result of such an aggregation is a single set of fertil'aer quantity- price data

in which each of the representative households strata exert an influence proportional to
the total quantity of fertilizers used by households of that stratum.
In pr’mciple, given a set of fertilizer and rice product prices, fertil'uer use can be
extended to the level where marginal return equals marg'mal cost. These levels of
fertilizer application are commonly known as “ optimum levels” of use. The aggregate
likely leveh of consumption, if all the farmers fertilize their entire crop area by optimum
levels of fertil’ner use, represent upper limit of the effective demand. However, in
practice, these optimum rates of application are discounted by farmers on account of
uncertainties of returns due to several climatic, social, economic and availability factors;
and the level of effective demand is largely determined by the irrrpact of these factors on
farmers’ decisions to use fertilizers in the context of prevailing situation.

The derived demand curves, at best, can only be regarded as short run estimates

74
duetouncertainty aboutfutureprices,technologieeand institufional constraintwhich

may be imposed on the system. Even in the short run, farm managers’ decisions may
vary substantialy from the actions predicted by the linear programming modeh of the
farm firm. In particular, different subjective estimates of managers in relation to crop
yieldsandprices,anddifferent attitudestoriskmay resultinfarmers’ actualdecisions
differing, somewhat markedly, from those indicated as optimal. Therefore, our basic
model will be modified to incorporate the risk component. This procedure is discussed
in subsection 4-5-5. The following subsection describes the cost structure in
transporting fertilizer.
4-5-4- Import Parity Prices for Fertilizer

In order to check whether or not ON can support a private fertil'uer distribution
system, we need to estimate the potential effective demand for fertilizer at the ON by
using or plugging into the model the import parity price of fertilizer at the ON. The
calculation of the import parity price of fertilizer might also reveal possible areas of
improvement in the marketing margin resulting from fertilizer import. The marketing
margin is the difference between import or ex-factory price and the retail farm price. It
consist of wholesalers’ commissions, retailers’ commissions, transportation cost,
storage cost, insurance, interest on stocks and facilities, and other overhead cost. It is
possible to determine the share of each of the component, which comprise the
marketing margin, and to determine the factors responsible for a high marketing
margin . The retail price of fertilizer can be lowered by reducing the marketing margin

through an increase in marketing efficiency. It will also be potible to derive the

75
nominal protection coefficient (NPC) and the nominal effective protection coefficient
(NEPC) from the import parity prices of fertilizer. From the standpoint of farm
’mcentives, it is important to determine the extent to which farmers in the ON are stil
protected after the devaluation. Protection implies that domestic producers of a
commodity i can be inefficient relative to foreign producers.

Thenominalprotectioncoefficient(NPC) isequaltotheratioofthedomestic

price of a commodity i to it border price using the official exchange rate (Sadoulet at al.,

1995):

NPCi =Pg/Pg,

fleas,
if NPCi is greater than 1, producers are protected and consumers taxed,

if NPCi is less than 1, producers are taxed and consumers subsidized, and
if NPCi is equal 1, the structure of protection is neutral.
The effective protection coefficient (EPC) can be estimated for a commodity I

(i.e. rice) at the farm level to capture the net effect of distortions such as input subsidies
(Sadoulet et al., 1995). The nominal EPC is: NEPCi =Vaid/Vaib where Vai is the value
added (or return) on primary factors in the production of I measured at domestic price
(40 and at borders price (b) using the official exchange rate, hence the term nominal.
IfNEPCi >1 or >0, domestic producers of r' are directly protected. The return on
their resources is higher than it would be if border prices prevailed, creating incentives
to increase the production of this commodity.
If NEPCi <1, domestic producers of i are disprotected, price distortion give them

disincentives in the production of i and they can remain in the production of i only if

76
they are more efficient than foreign producers.

If NEPC i=1, the structure of prices is neutral on incentives.

T‘wotypesoffertilizersareusedbyricefarmersintheON: ureaandammonium
phosphate. Urea is imported from outide the continent. Ammonium phosphate is
imported fiom Senegal’s chemical industry (ICS) or from Hydrochem-CI in Céte
D’Ivoire. While the brrlk of Mali’s imported fertilizer goes through the Dakar-Bamako
region route by train, or the Abidjan-Ségou-region route by trrrck, a smal portion
comes from Nigeria and Niger. The main cost items in importing fertil’ner are
international and domestic transportation costs, handling and insurance cost. More
explicitly, the calculation of the import parity price of fertil’uer includes the following
cost items:

1) The FOB price at the export point, plus freight and insurance, plus
unloading at import dock, corresponding to the HANDLING price,

2) The HANDLING price plus tariffs, minus subsidies, plus port charges,
transport and marketing to the relevant market or project boundary
correspond to the import price at central market or project boundary
(IPPCM or IPPPB). Tariffs and subsidies are excluded in economic analysis.

3) The IPPCM or IPPPB plus the local transport, storage cost, etc, between
the market and the farm gate, corresponding to the import parity price at the
farm gate.

In this study, the main central market or project boundary is Niono, which

represent the point of entry to the intensive system (Retail zone), semi-intensive system

77

(Arpon zones located next to and far away from Niono) and non-restored area far from
Niono.
4-5-5- Modflication Of the Basic Model to Incorporate Risk Component

Assaidearlier,farmersintheONarefacingtheriskofincomevarinbility
resulting from weather and price variabilities. Assuming that farmers in the ON are
facing the same market price conditions, the main difference among them is weather
and the degree of water control. Indeed, improved varieties of rice cultivation in the
semi-intensive system and non-restored area are more affected by the distribution and
level of rainfal than are those in the intensive system, where water is relatively wel
controfled.

When linear programming is used in a decision support role, risk can sometimes
be assessed outide the formal framework of the model and the farm plan indicated as
optimal may be adjusted in a subjective way (Dent et al., 1986). It is also possible to
’mcorporate risk formally into the planning framework using linear or quadratic
programming methods. If a quadratic risk programming study is to be undertaken,
then it is necessary to estimate both the expected gross margins for each activity and also
the varhnce of these returns. In addition, estimates must be made of the covariances
between al activity returns; that is, the extent to which the returns for different
activities vary together. The objective function in the quadratic risk programming is
specified in non-linear quadratic form, accommodating the variance and covariance of
the Cj values. The solution is then in terms of a plot of the expected values of Z, denoted

E(Z), against it variance V(Z) (which is taken as a measure of the risk faced). Choice of

78
the optimal solution then depends on matching the feasible set of [E(Z), V(Z)] values

againstthefarmer’spreferencessoastochoosethepair,whichgiveshimorherthe
greatest utility.

Quadratic risk programming is demanding of both data and computing
resources. Indeed, the considerable data requirement, exacerbated by a lack of widely
avaiahle trouble-free solution algorithms, have so far restricted the application of
quadratic risk programming as a practical decision aid in solving applied management
problems (Dent et al., 1986). Because of this, a variety of linear approaches to
accommodate risk have been developed. Such modifications include simplex linear risk
programming, minimization of total absolute deviations (MOTAD) programming and a
variety of extensions of MOTAD. In particular, MOTAD has the attraction of being a
linear approximation to quadratic risk programming (Hazell, 1971).

The income variability will be taken into account in the semi-intensive system
and non—restored area by developing a MOTAD model accommodating the constraint of
risk via the incorporation of possible states of nature and their probability of occurrence.

The conventional MOTAD model as developed by Hazell in 1971 is as follows:

1) Minimize: 3.4 = : (Y,+ + Y,')
if 1:]

such that for each year,

where;

79

mike. —G,)X, -Y: +Y: = 0

M

20,), = E
i=1
gangs
1=I

sA = total absolute deviations of farm income over al years

s = the number of years over which income is sampled,

A = the mean annual absolute deviation of farm income,

X, = the level of the jth activity or enterprise,

Q - the gross margin (ie, gross return over operating cost)

forthejtlr activity inyeart,

G, = the annual gross margin for the jth activity,

Cg—G, = the gross margin deviation for the j... activity in year t,

E = expected net income set equal to some specified level,

A, = the input-output coefficient showing the unit of the ith
input required by the jth activity,

b. = the quantity of each resource,

Y: = an accounting enterprise in the LP matrix entering the

MOTAD solution when the total income deviations for a

80
particular year t are positive,
Y.’ - an accounting enterprise in the LP matrix entering the
MOTAD solution when the total income deviations for a

particular year t are negative, with

Y.- or Y: =

 

 

1230,. - G,)X,.

By means of equality constraint, the value of expected net income (E) is set at
specflied increment, from zero up to the linear programming profit maximizing level to
develop a series of solutions. Plotting expected total gross margin against total minimum
absolute deviation yields the efficiency frontier. Each solution is efficient in the sense
that no other enterprise mix will result in less income variation at the specified level of
income.

In the semi-intensive and non-restored area in the ON, the farm household want
to obtain a minimum income from rice and vegetable crops production to cover part of
it expenses. To include this objective in the structure of the LP model, additional rows
are needed. These rows contain the deviations from the mean income of rice (produced
on ON plot) and vegetable crops. The mean income for each crop is calculated by
weighting the income from one hectare of that crop associated with each state of nature
by the probability of occurrence or cumulative distribution or frequency of that state of
nature. The probability of occurrence of any state of nature is calculated from historical

annual rainfall data from which it is possible to determine a relatively good, average and

81

badyearsirrtheregionalcontext. Itisassumedthatincomeperhectare'noneofthe
uncertainvariablessinceboththeyieldandthepriceofriceandvegetablecropsmay
fluctuate Table 6 below illustrates the way the minimum income constraint wfl
includedinthemodel.

Table 6: Incorporation of the Minimum Level of Income.

 

 

 

 

 

 

 

Level Relationship Activities
Rice Vegetable crops Z1 Z2 Z3
l (he) (he)
Expected Total Gross + Equal Y E
arsi-O'GM)
Yearl GM tie + Greater than or + + 1
Equal
Year2 GM tie + Greater than or + + 1
Equal
Year3 GM tie + Greater than or + + 1
Equal
C Min 1 1 1

 

 

 

 

 

 

 

 

 

 

Source: Front the author.

On Table 6 above, the expected total gross margin (from rice produced on ON
plot and vegetable crops) will be set at a specific level by means ofan equality constraint.
The three gross margin tie rows are introduced for three representative types of years
(Good, average, and bad) based on the levels of rainfall received. Also for each year, a
shortfal activity (reflecting the amount by which the gross margin fails to reach it

expected level) is added. Shortfall activities are labeled Z1 to Z3. Each of the gross

82
margin ties stipulate that the total gross margin plus shortfall for the year be not less

than the specified level oftotal gross margin. The average of the gross margins for the
three~yeartypes istheexpectedgrossmarginintheexpectedTGMrow. Sincethe
expectedgrossmarginforanyactivityisthemean(X)acrossallindividualyears,the
sunrefdeviationsabovethismeanwiflequalthesum ofdeviationsbelowit(Dent, 1986).
Hence, minim—ng the sum of shortfalls minimizes the sum of deviations both above
and below the expected gross margin. The C row on the table states the objective of the
head of household as minimization of the sum of shortfalls, that is, of the sum of negative
deviations from the expected total gross margin. The choice of this objective function
ensures that the sum of the shortfall will be as small as possible, that is, that the most
stableplanintermsoftotalgrossmarginwillbeselected.

The farm household want to produce on average a large share of the coarse
grains (millet, sorghum, maize, rice) and root crops to cover it needs. To include these
considerations in the analysis, additional rows will be introduced in the LP matrix in a
way simflar to that for the constraint on the level of income. These rows contain the
deviations from the mean during the worst possible states of nature for the yields of food
crops produced for home consumption. The mean yield is calculated by weighting the
yield associated with each state of nature (good, average and had years) by the
probability of occurrence of that state of nature. It is assumed that the yields of food
crops are the uncertain variables. The yields are expressed in thousand calories to allow
for the same unit across an crops. Table 6 below illustrates the way the minimum level

of foods production for self-sufficiency will be accounted for. On the table, the Zs are

83
shortfal activities.

It was estimated (DNSI, 1988-89) that the per capita consumption of cereals per
year in Mali is 203.81 kg, rounded to 204 kg. This figure was lower than the
consumption norm of 212.4 kg reported by Steffen in 1995, implying that some other
sources were needed to cover the shortfalls. The shortfalls could be covered up from food
aidorbybuyingcerealsfromthemarketbringingabouttheissueoffoodsecurity
discussed by Sadoulet and De Janvry in 1995 . The definition of food security provided
by Sadoulet and De Janvry was as follows: Food security menus access by an people at
al time to food sufficient for a healthy life.

It is clear from the above definition that if food security cannot be obtained
through domestic or own production the gap should be cover up from market or from
outide the county. Thus assuming that an average person in the ON must consume
212.4 kg of cereals per year to meet the minimum requirement of 2300 calories per
person per day, an average household of 10 persons must consume in minimum 2124 kg
per year. Since households in the ON are more likely to fall into transitory food
insecurity because of many sources of risk (e.g. fluctuation in income, production and
prices) during average and had years, food security must be tackled with the broad
concept of food security in covering up the shortfalls (ZZ and Z3) defined in table 7
below. The level of expected yield during the average and had years can be varied
within an acceptable range in order to analyze the magnitude of the shortfall in each
scenario. Such analysis can reveal the need to rely on other possible sources (i.e. market,

food aid) that may help covering the shortfalls in a risky situation.

84
Table 7: Incorporation Of the Minimum Level Of Food Crops Self-Sufficiency.

 

 

 

 

 

 

 

 

 

 

 

 

 

Expecteineld Levelllelation Activities
s

MrlX X X X Vegetable Zl Z2 Z3

let a
swarm + r :Y" 'X— Y 3? Y I?
Yearltie + GorE + + + 1
Year2tie + GorE + + + 1
Year3tic + GorE + + + + 1
CMin 1 1 1

 

 

 

 

 

 

 

 

Source: Front the author

NB: The plus (+) signs on the table correspond to positive levels of yield. G and E stand
for “Greater” and “equal”, respectively

85

CHAPTER 5 - DATA NEEDS AND PROPOSED DATA

COLLECTION STRATEGY

5-1- Data Needs

In each rice production system in the ON, we need to collect data in order to
derive enterprise budget and the input/output coefficient for each enterprise under
consideration. Enterprise budgets constitute the key building blocks of the LP model.
Detailed information needs to be collected on the sampled households’ labor force, labor
h'ned in mandays and wages, off-farm employment, the various input including
fertiizers and their respective cost, the market prices of crops, the sources and amount
of credit and the rates of interest charged. We also need data to derive the import parity
prices of fertilizer. The type of the needed data in deriving the import parity prices for
fertilizer or any commodity has been described in section 4-5-4. These data will come
both from primary and secondary sources.

Data requirement for both the conventional LP model and the MOTAD model
are the same with respect to Input-output coefficient and expected cost and returns.
However, a time series of gross margins (gross returns over operating cost) and crop
yields are required for each enterprise to develop the income and yield deviations for the
MOTAD LP matrix. The determination of the distributions of gross margins over many

years involves the determination of the distributions of several random variables. The

86
most analytical approach to specifying revenue distributions is to begin with the

underlyingdttributionsoftheuncertain component suchasyieldandprice(Dilon
and Hardaker, 1977). Usually, only means, variances, and covariances are required so
that assessment need not be too demanding.

LetGM -riceenterprisegrossmargininCFAF/ha

Y - rice yield in in kg/ha

P = rice price in CFAF/kg

VC 8 variable costs of rice in CFAf/ha.
Let assumes that VC is known with certainty. So that they do not influence the variance
of the gross margin GM which given approximately by:

V(GM'z’EfflEffD V00} + {15090319 V(H} + flEmEmCWfYfl}
V(GM), V0) and WE are variances of gross margin, yield and price respectively
Cov(Y,fl is the covariance between the yield and the price.
E(H, E(I) are the expected price and yield, respectively.

The expected gross margin is derived as follows:

E(GM)= E(DEm + Cov(Y,H -VC
Even after having gone through the problem of estimating revenue d'ttributions for
several enterprises, the analyst is still left with the difficulty of establishing the
appropriate correlations between enterprises returns. Dillon and Hardaker (1977)
reported that one of the solutions to this problem requires knowledge of at least one

formula for the parameters of a conditional distribution for one variate of a bivariate

normal distribution. UK] and X2 are jointly normally distributed with means E(X 1)

87

and EM); standard deviations SD] and SDz; and correlation Carry; the conditional
distribution ofXI, given XfX2*, is characterized by mean and variance

Em‘Xz-Xfi = E(XI) + C0fi12(SDI/SDzllX2 *-E(X2)]

V(X 1W2 I"X29 = Corr] *Corrfll-Corrz] “Corral

Attention should be given to the above possible correlation relationships when two or
more enterprise gross margins in this proposed study are found to be correlated

through years. Appendix A present the linear programming tableau including the
consumption activity while appendix B provides the programming matrix including the

possible correlation between rice enterprise and the other enterprises in the ON area.

88
5-2- Data Collection Strategy

Thedatausedinthisstudywflbeobtainedfrom both primaryandsecondary
sources. Secondary data will be collected by consulting relevant report written about
the ON. Primary data will be collected through a survey method by means of
questionnaires and personal interview. Data from the survey will be used to undertake
the cost and returns analysis and to implement the programming exercises. The
colection of the field data will be done by trained enumerators under the supervision of
the researcher. The collection of secondary data, the implementation of the
reconnaissance survey and the implementation of the interviews will be canied out by
the researcher with the help of ON extension workers.
5-3- Sampling Procedure and Sample Size

In deriving demand estimates by linear programming, it wifl be prohibitively
expensive and' time-consuming to estimate the demand schedules for each farm
household in a region and then to derive the regional demand function by summing the
individual firm functions. An alternative is to select a sample of representative farms in
order to guide planning on individual farms. Barnard and Nix (1973) suggested that in
areas where there ’n reasonable homogeneity in at least some of the major resources -
particulariy with respect to natural factors, such as soil type, topography and climate —
linear programming can be used to obtain solutions to ‘modal’ or ‘representative’ farm
situations. In this way some of the benefits of comprehensive computer phoning can be
made available to a greater number of farmers in an area than would be the case with

the individual programming of farms. Since farms are likely to display considerable

89
variation around a particular modal situation (when account is taken of both

quantitative and qualitative aspect of farm resources), more than one model 'n
requhed if differences in factors such as farm size, the number of workers and the
avaiability of building are to be accommodated. Therefore, we need a sample of
representative farm households, each supposed to be characteristic of a larger group of
farm in the region. Each representative farm demand curve is weighted by the number
of farms in it stratum, and the weighted demand curves are then summed to derive the
regional estimate.

Stoval (1966) discussed three sources of error that may bias the demand
schedule estimated from a linear programming model and a sample of representative
households:

(i) specification error;
(ii) sampling error; and

(iii) aggregation error.

Specification error arises when the programming model fails to accurately
describe the conditions faced, the derived objectives, and the resulting decisions being
made by the firm.

Sampling error arises when the distribution of the variables of interest within the
sample differs from the distribution of those variables within the population. That is,
when a sample is taken from a population, it will not be possible to know precisely the
value of any population parameter, such as the mean or proportion. Any point estimate

will be in error. The problem of reducing and measuring sampling error are tackled by

90
stat'ntical procedures and sampling theory is sufficiently developed so that sampling
ratescanbesetsoastoholdsamplingerrortoadesired level.

Frick and Andrews (1967) have defined aggregation error as “.the difference
between the area supply (demand) function as developed from the summation of linear
programming solutions for each individual farm in the area and the summation from a
smal number of typical or benchmark farms.” In other words, aggregation bias occurs
when the result obtained by scaling up solutions from one or more representative farms
are different from the result that would be obtained if it were possible to solve for each
'mdividual farm in the target population and add the result (Upton and Dixon, 1994).

Aggregation error or bias is brought about by the use of averages or other
measures of central tendency in synthesizing a model unit from survey data. According
to Collinson (1972), aggregation bias is the same phenomenon noted in supply response
work; in this case interfarm differences in timing create different peak requirement on
particular farms, which are damaged when averaged - and peaks on one farm are offset
by relatively slack periods on another, so that the whole labor profile is flattened. An
alternative to straightforward averaging is the construction of the model from
component, each of which is sampled, perhaps independently, for the population under
investigation (Collinson, 1972).. Such construction reduces the aggregation effect and
highlight the decisions required on the inclusion of specialized activities, which are
recorded only on a proportion of the sample units. The component of a detailed labor
profile are the activities identified in the system, the acreage, calendar, operational

sequence, and rates of work for each operation. Improving the simulation of the profile

91

requires attention to the timing of crop operations. Construction reduces the
aggregation bias by enumerating timing, the center date for each discrete operation, as
a component. The steps involved in a profile construction is described by thoroughly
described by Collinson (1972).

The necessary conditions for selecting representative farms to minimize
aggregation error within a given budget are still undefined in a general sense. However,
Miller (1967) has demonstrated that two criteria are useful to control aggregation error
in an empirical situation. Miller recommended that:

(1) farms be grouped on the basis of what is the most limiting resource in

the production process; and

(2) that farms with similar patterns of product response to price change

be grouped together.

Upton and Dixon (1994) reported that some bias can be avoided by careful
grouping of the population of farms prior to selecting the representative cases.
They suggested that farms that produce similar products in similar climatic zones be
grouped together as a first step, and then use the statistical technique of cluster
analysis to identify sub-categories based on resource ratios. Furthermore, they
reported that when the models are of representative farms and results are to be
scaled up to describe behavior of a population of farms, it becomes important to
take some account of aggregate effect on prices by linking MP models to market
models in a recursive fashion or by separate estimation of price changes and

consequent use of the prices for the farm model. Selected input and output prices in

92

representative models can be varied parametrically to derive demand and supply
schedules, which can then be scaled up across categories as appropriate. Given other
information about the supply of farm input or the demand for output, market
equilibria can be deduced.

Thus, for this proposed study, a sample of a representative group of farm
households within each rice production system can help reduce the problem of
' aggregation error and serve the purpose of this study. The techniques of sampling based
on field experiences in conceptualizing and implementing policy- relevant studies have
been thoroughly discussed by Teff‘t etal. (1990).

Theotherimportantsourceoferrorisnon-samplingerrors. Thistypeofenoris
unconnected with the kind of sampling procedure used and tend to be greater than
sampling errors (Liedholm, 1998, personal communication). Indeed such enors could
just as wel arise if a complete census of the population were taken. In any particular
survey, the potential for non-sampling error exist at a number of places. Newbold
(1991) provided some examples as follows:

1. The population actually sampled is not the relevant one.
2. Survey subject may give inaccurate or dishonest answers.

3. Non-response.

(4)-Data entry or processing errors
In addition to the three sources of errors provided by Newbold, other
types of non-sampling errors arises during data transcription and data entry stages.

There t the possibility of an error connected with enumerator errors. Tefff et aL (1990)

argued that minimizing non-sampling errors during the data processing stage
demands consideration of several other aspect, such as the design of questionnaires,
hiring knowledgeable computer personnel, planning the type of analysis to be
completed, and selecting hardware and software that satisfy researcher needs
throughout the implementation of any project.

There is no general procedure for identifying and analyzing non-
sampling errors (Newbold, 1991). The main prescription is that the investigator
take care in such matters as a) identifying the relevant population, b) designing the
questionnaire, and c) dealing with non-response in order to minimize their
significance. Tefff et al. (1990) argued that minimizing non-sampling errors depends
on well-defined concepts, adequate operational definitions, accurate transhrtions,
and simple and easily understandable questionnaire formats. In addition,
researchers must carefully train enumerators and supervisors, provide them with
explicit instructions and sufficient logistical support, obtain the cooperation of
village leaders and respondents, and initiate data verification and analysis early in
the study.

In regard to sample size, Yang (1965) warned that:”..in no
circumstances should the research worker choose a sample larger than his financial
and personal resources..”. Moreover, Yang stressed that when 20 farms are selected
from the same stratum for cross-tabulation, the addition of more farms will not

seriously change the results..., that roughly 20 farms should be included in each of

the classes in order to make reliable comparisons.” Yet for this study, 30 farmers
will be selected in each system in order to anticipate the possible lack of cooperation
of some farmers and other problems associated with farm management
investigations. Resources, activities, and constraints pertinent to the ON area will be
collected in a survey of 90 households during a cropping season. Linear
programming models will then be developed to present the major household types in
the ON area. Parametric solutions will be obtained for those resources limiting in
the initial optimal plan for each model, as the scarce resources influence the MVP
and hence the demand for fertilizer. The variability of income and yield risks
farmers faced will be taken into account by developing risk-programming models

for the study area.

Appendix A:

Programing Matrix Including the Different Activities in the ON area

95

 

Lanes (IIFann (llfann (llfann (IIann
J t .t

A.'._...

Mitten! Mititll erllll Lila! Lllfll Lita! LIICI

II
(his

lefll
thlcul

"hon “he! "he!
can out on.
‘I. ,t A _

J

'4‘.

"it‘ll tilt!
AL.‘ 11

---.... .__

.. and“ o

tlre‘ln Lanes Lahar tenor Lanes Lanes Labor
lint “heels Nicola unnlhr anoint «koala “heels Nicola

('l ('l

('l

(‘l f°l (°) ('l

('l ('l

('l

('l

0 ('l

f

(i

(i

(i

(i

(i

C.)

('l

('l

l'l

('l

('l

('l

('l

('l

A
O
v
A
0
so
A
o
v
A
0
up
A
n
v
a A
0 O
iv v
A an
O O
v v
A A
O 0
v v
A mm
D O
v v
am A
O 0
v v
a A
O 0
our v
A
n
v
A
0
or
A
O
v
4‘
U
f'
U
T‘
V
f‘
v
1‘
it
<‘
U
‘P
a—
1‘
v
an A a A
O O O 0
v v v m!
A A A an all A A an
0 O O O O O O 0
it or v v m— - in v
A a
O 0
us mm
A an
O 0
v v
A A
O n
a— v
A A
O 0
up in
A an
O O
‘— v
a A A a A A a n—
. O O O O O 0 0
v v v v m! V v v
A at an A as A A a
O O O O O 9 0 n
U U n— - v V V mm

film!!! ilrlll illililll Ill

Lahoeroeroerer
. June -.--

.J

Lmor 0'!me 0|!me ONE-m 005mm

1.“
llreomn Modem ltlrarlmn

J

LOG

"illiil
three
era
in Guam-

 

M“ I.“ W l...
d ml ml WC!“
l . .-

rem
C!

W

L“

‘ A 5'“... - _

L“

I... w W
“h “It “Us Mb “h “It Mb

I...

iilii””
lllulll
Ilse

 

f‘l ('l ('l

(’l ('l

i l‘l

f

M ii

('l

(i

f-l

('l

 

('l l‘)

f') ('l

'l

rl lsilll

A

a
0

iii if. ii: lr: 13.11:

l'l

l‘l

P)

r..-
ru menu

can.

if

 

“I:
(‘l l“) ('l ('l

('l (‘l (‘l l')
H (l (l (l

8 A
0
~—

1 :2. s; m u m hi hi

 

 

ii:
iii;

0)

P) ('l

P)

(‘l

('l ('l

l'l

('l

 

 

.
4
l t
e

‘ '1‘

U

a

O
H

g; E
5!
ii
is

w'o'fi-s-

(‘l

(‘l

 

(or ' to)

('l ('l

......

 

p
lii
I. q

I!
(an.
I:

:28
gm
‘0:

9.520 9328
9.13 also

tutu

23:6

1:28

:33

12.20

8x3 Ilsa och.- nil in
In al.-Inna 85 i {In all
B»! i

no?! 9528 :36

 

 

m Caron-m In“. krona. Barents. harm horror-em layman.
Fr F

If“

“lid!
“in.
'1 L -. ‘ .
0 0 0

Ft- From From
Inform!
m J

I ... .
._ -m.‘ AM-

From Fr

or
m lrrfnrmd lain-mi mmd Inform!
was was Imus Soiree mes Sauce

Frsmr

r-
mun
trance

laE

 

 

 

(0) 105

 

('l

 

 

 

('l

(-l

H

‘fl

5

"‘-

P)

P)

(‘l

(‘l

(‘l

(“l

('l

P)

('l

P)

P)

('l

f‘)

f')

('l

('l

° i

f2

APPENDIX B:
Programming Matrix Including the Possible Conealtion Realtionships between Rice

and other Enterprises in the ON area.

Programmlng Matrix: Possible Correlations between Rice and Other Enterprises In the ON Area

 

Ricelnslde ON Rico Outsldo ON Onion Tomato Garlic Pepper Sweet Potato must/Sorghum Maize Total Supply

 

 

 

Expected Gross Revenue .- FCFA (*l (t) (+) (t) (+) (*l (t) (+) (+) (t)

Variable COS! - - FCFA (+) (+) (+) (+) (+) (+) (*l (t) (t) if)

Expected Gmss Margin FCFA (+) (+) (+) (+) M (f) (t) (t) M (+)

Standard Deviation Of Gross Margin FCFA (+) (+) (+) (+) (+) (+) (t) (+) (+) (t)

Coefficient Of Variation _ '

Risk Adjusted Gross Margin - FCFA (+) (+) (+) (+) (+) (+) (+) (+) (+) (+)

Resource Use ‘ A . - ’

’Land Inside 0N ' HI (+) (+) (+) (+) (+) (+) (+) . (+) (+) (+) (+)
‘_ [and Outside ON ._ ‘ Ha (+) (+) (+) (+) (+) (+) (+) ,_ (+) (+) (+) (+)
' .Family Labor 1 3 - ' “and!” (+) (+) (+) (+) (+) (+) (+) (+) (+) (+) (+)
— Hired Labor , ‘ ' 11¢th (+) (+) (+) (+) (+) (+) (+) (+) (+) (+) (+)

Watkins Capital ' . ' - ’ fl (+) (+) (+) (+) (+) (+) (+) (+) (+) (+) (_+)
. Correlation Matrix — ~ -

.Riceinsr’deON ' " ' - V ' g l _

.‘RiceOutsideON . g ' * (+) . 1
Onion . g - ‘ _ (+) .' (+) l
Tomato ‘ -‘ ' ' o (+) (+) (+) l

-‘ Garlic _ J ' . (+) (+) (+) (+) 1

Pepper. y g . , -. (+) (+) (+) (+) (+) I
*_ Sweet Potato. , -_ ' 7 g o g. (+) (+) (+) (+) (+) (+) 1

Milleu‘Sorghum - (+) (+) (+) (+) (+) (+) (+) 1

. W29 ' ~ - -' __ (+) ‘ A (+) (+) (+) (+) (+) (+) (+) 1

97

BIBLIOGRAPHY
Barry, A.W., 1994. Comparative Advantage, Trade Flows And Prospect For

Regional Agricultural Market Integration In West Africa:
The case Of Cote D’Ivoire And Mali. Unpublished Dissertation for the Degree of
PhD Michigan State University.
Beneke, R. and Winterboer., 1973. Linear programming application to
agriculture, Ames, Iowa State University press
Barnard, C. S. and Nix, J. S.,l973. Farm Planning and Control, Cambridge University
Press, Cambridge, London. New York. Melbourne.
Boughton, D. et al, 1994. Analyzing the Impact of Structural Adjustment
on Commodity Subsectors. Currency Devaluation and the Maize
Subsector in Mali
Bumb, B.L. and Teboh, J.F., 1992. The Policy Environment and Fertilizer
Sector Development in Mali. An Assessment.
Camara, A.,1988. Economic Analysis of Traditional and Improved Farming
Systems and Optimal Farm Plans for Peasant in the OHV Zone, Mali.
Unpublished Masters Thesis, Texas Tech. University, 1988.

Cisse, Y. et al, 1993. Etude Recherche Developpement h l’Office du Niger.

Rapport d’etude non public, Ministers Du Development Rurale et de
L’environnement, institut D’economie Rurale.

Cisse, Y., 1987. Resource Use and Productivities. A Comparative Study of

98
Intensive and Non-Intensive Farming Systems in “Mali-Sud”
Rural Development project. Unpublished M.sc.Thesis,
Ahmadu Belo University, Zaria (Nigeria).
Clayton, E.S., 1964. Agrarian Development in Peasant Economics.
Some Lessons from Kenya. Perganon Press.
Coll'mson, MP. Farm Management in Peasant Agricuh'ture.
New York, Prager, 1972.
Coulibaly, O., 1993. Farm Level Potential Impact of Soil Fertilization
and Water Retention Techniques on Sorghum Based Cropping

SystemsinMali.

Coulibaly, B.S., Sanogo, O., Marika, D., 1994. Coht dc Production du
Paddy h l’Office du Niger. Atelier Régional sur l’impact de la
Devaluation du FCFA sur les Revenus et la Securité Alimentaire
en Afrique de l’Ouest.

Del, V. et al, 1995. Les Premiers Effets de la Devaluation sur les Filiéres Riz
Irrigué au Sahel. Le Cas du Mali. IER-ClRAD. Ministére de la
Cooperation. République Francoise.

Dembele, N.N., 1996. “Implication of Market Reform for Fertility
Technology Development and Implementation in Sub-Saharan
Africa with Special Reference to the Soil Fertility Management
Initiative in Burkina Faso.” Unpublished study report,

International Development Center -Africa

(IFDC-Africa). Soil Fertility Management Unit (SFMU), Burkina Faso.
Dembele, N.N., 1994. Economic Analysis of Trader’s Response to Cereah
Market Reforms in Mali. Unpublished Dissertation for the Degree of Ph.D.,
Michigan State University. Department of Agricultural Economics.
Deme, M., 1993. La Filiere Riz au Mali. Situation- Problemes-Perspectives
Rapport Pr'mciml-
Demeke, M., Said, A., and Jayne, TS, 1996.” Relationship Between Input
Markets and Grain Market Performance in Ethiopia.
”.Paper Presented at the Grain Market Research Project
Discussion Forum”, Sodere, Ethiopia. Ministry of Economic

Development and Cooperation

Dent, 1.3., 1986. Farm Planning with Linear Programming: Concepts

& Practice.
Dillon, 11.. And Hardaker R. Agricultural Decision Analysis (1977),

Iowa State University Press, Ames.

D'nrra, 83,1994. The Role of Small Rice Mills in the Rice Subsector of the
Ollicc du Niger, Mali. Unpublished Master thesis.
Michigan State University. Department of Agricultural Economics.
Dione, J., 1989. Informing Food Security Policy in Mali. Interactions
Between Technology, Institutions and Market Reforms.

Unpublished Dissertation for a Degree of Ph.D. Michigan State University.

Department of Agricultural Economics.

100
Dione, J., at al., 1996. ” Adjustment Structurel, Politiques Economiques et

Sécurité Alimentaire Au Sahel. ” Contribution au Forum International
du 20cm Anniversaire de l’Institut Du Sahel.”
Dione, J., er al., 1996. “ Contraintes a I’Investissement pour la Productivite
et la Competitivite des Systemes Agricoles et Alimentaires en Afrique
de I’Onest: Question Trans-Filiéres et Trans-Nationales.”
DNSI: Direction N ationale De la Statistic Et De L’InformatiqueEnquete Budget
Consommation, 1988-1989.
Diagana, N.B., 1995. ” Devaluation du Franc CFA et Decisions de
Production Agricole. Une Analyse des Impact de la Devaluation
du Franc CFA sur les Choix de Culture et de Technologies de Production
par les Ménages Ruraux du Bassin Araclridier du Sénégal.”
Communication Préparée pour I’Atélier Regional CILSS/PRISAS
sur le Theme “.Impact de la Devaluation du Franc Cfa sur les Revenues

et la Sécurité Alimentaire en Afrique de l’Ouest.” Term 2. Bamako, Mali.

Etuk, EG. ,1982. A Comparative Study of Optimal Farm Plans for Small Farms
with Traditional and New Technology in Zaria Area of Northern
Nigeria. Department of Agricultural Econmics and Rural Sociology,
Ahmadu Bello University.

Falusi, A., 1973. Economics of Fertilizer Distribution And Use in Nigeria.

A Thesis Presented to the Faculty of the Graduate Scholl of Cornell

101
University for the Degree of Doctor of Philosophy.

Flinn, J.C., 1969. ” The Demand For Irrigation Water In An Intensive
Irrigation Area.” University of New England.

Frick, GE and Andrews, RA.” Aggregation Bias and Four Methods of Summing Farm

Supply Functions. “ Journal of Farm Economics, 1967.

Gerner, H., Jager, de. M., Tboh, F.J,Bumb., and Dembele, 1996. “Economic
Policies and Fertilizer Market Development.” In Restoring and
Maintaining the Productivity of West African Soils.

Gittinger, J.P., 1982. Economic Analysis 0] Agricultural Projects. Second
Edition. The Johns Hopkins University Press, Baltimore And London.

Hazell, RR, 1971. “ A Linear Alternative to Quadratic and Semivariance
Programming for Farm Planning under Uncertainty.”

American Journal of Agricultural Economics, vol. 53, No.1.

Heady, E0. and Diflon, J. L. (1961). Agricultural Production Functions.
Iowa State University Press, Ames, Iowa.

Henao, J. et al, 1982. Fertilizer Policy Research Program for Tropical
Africa. Agronomic Potential of Fertilizer Use in Mali.

Heyer, J., 1971. ” A Linear Programming Analysis of Constraints on Peasant
Farms in Kenya. ” Food Research Institute Studies in Agricultural

Economies, Trade and Development, vol. X, No.1.

[ER (Institut D’Econonrie Rurale). Enquétes 1989/90.

IFDC_Internatioual Fertilizer Development Center,

102
“.1995 Global Fertilizer Perspective 1980-2000.

The Challenges in Structural Transformation.”

Kamuanga, M., 1982. Farm Level Study of the Rice Production System at the
Wee du Niger in Mali. An Economic Analysis. A Dissertation for the
Degree of Ph.D., Michigan State University. Department of
Agricultural Economics.

Kely, A.V., 1988. Factors'Alfecting the Demand for fertilizer in Senegal’s
Peanut Basin. A Dissertation for the Degree of Ph.D.

Michigan State University. Department of Agricultural Economics.

Kottke, M.” The Anatomy of a Step Supply Function.”

Journal of Farm Economics, 1967.

Longman Dictionary of contemporary English, New Edition, Longman Group
UK Limited 1987 (Secon Edition). Page 1055.
Low, A.R., 1974. ” Linear Programming as a Tool in Innovation Selection
for Farms in S.E. Ghana, in Planning in Low Income Countries.”
A Symposium Development Study No.14, Reading. Factors Affecting
Demand And Estimation of Future Demand. Baroda (India.)
Martin, F., 1988. Food Security And Comparative Advantage In Senegal:
A Micro-Micro Approach. Dissertation for the Degree of Ph.D.
Michigan State University. Department of Agricultural Economics.
Ma'iga, S.A., 1993.” Revue Bibliographique et Description Sommaire de Mécanismes

de Fonctionnement du Marché des Engrais.”

103
Ministere Du Developementr Rurale et de L’errvironnement.

Maiga,S.A.,l983. OptimumFarmPlansinIrfigatedAgricultureAcaseStudyof
the Niger River Development Project Office du Niger in the Republic of Mali.
Unpublished M.SC. Thes'u, Ahmadu Bello University, Zaria, Nigeria.

Maharaja, ME, 1975. Demand For Fertilizer. An Analysis of Factors
Affecting Demand and Estimation of Future Demand. Printed by
Nutan Printery, Baroda and 3A Associates Baroda.

McCarl ,B.A. and Spreeu T.H., 1980.” Price Endogenous Mathematical Programming
Models as a Tool for Sector Analysis.” American Journal Of agricultural
Economics. 62(1) pp 87-102. .

McIntire, J., 1985.” Constraints to Fertilizer Use in Sub-Saharan Africa.
Management of Nitrogen and Phosphorus Fertilizers in
Sub-Saharan Africa proceeding of a Symposium, Held in Lorne, Togo
March, 1985.

Mhrot, N.W., 1992. Distributional Effects of Currency Devaluation on Households
in Rwanda. An Application of Willingness to Paym Welfare Measure.
Unpublished Dissertation for the Degree of Ph.D.

Michigan State University. Department of Agricultural Economics.

Mfler, T.A.” Aggregation Error In Representative Farm In'near Programmin
Supply Estimates.” Unpublished Ph.D. Thesis, Iowa State University.,

Ames, 1967.

Mudahar, S.M., 1978.” Needed Information and Economic Analysis for

104
Fertilizer Policy Formulation.” Indian Journal of Agricultural
Economics, voLXXXIH, 33(3).
Mudahar, S.M. ” Fertilizer Problems and Policies In Sub-Saharan Africa
in Management of Nitrogen and Phosphorus Fertilizers In Sub-Saharan
Afria.” Proceedings of a Symposium, Held in Lorne, Togo,
March 25-28, 1985. Edited by A.UZO MOKWUNYE and
Paul LG.VLEK IFDC, Muscle Shoals, Alabama, USA.
Ngwira, A.N., 1994. The Role Of Dirnba Land And Small Scale Irrigation
InSmalholder Farmers’ Food Security I Target Motad
Mathematical ProgrammingDissertation .for the Degree of Ph.D.
Michigan State University. Department of Agricultural Economics.
Niang, A., 1980. A linear Programming Model of a Representative Sahelian
Farm. The Cases of the Cotton Zone in Mali and Peanut zone in
Sénégal. Ph.D.Thes'u, Purdue University.
Norman, D.W., 1973. “ Economic Analysis of Agricultural Production and
Labor Utilization Among the Haussa in the North of Nigeria,”
African Rural Employment Paper No.4, Department of Agricultural
Economics, Michigan State University.
Newbold, P., 1991. Statisties for Business and Economics. Third edition by
Prentice-Han, Inc.
Ogunfowora, O., 1972. Derived Demand, Product Supply and Farm Policy in the

North Central State of Nigeria. npublished Ph.D. Thesis, Agric. Econs.,

1 05
Iowa State University, Ames, Iowa.

Ogungbile, A.O., 1980. An Evaluation of Improved Solc~Crop Production
Technology or Small Farms in Northern Nigeria under Different Farm
Power Sources.lA Multi-Period Linear Programming Approach.
Unpublished Ph.D. Thesis, Iowa State University.
University of lbadan. Ibadan, Nigeria.

Roth, M. J. and Abott, RC, 1990.” Agricultural Price Policy, Food Aid and
Input Subsidy Reforms in Burkina Faso.” Journal of

Agricultural Economics vol. 41, no.3, sept.

Rae, AM, 1977. Crop Management Economics. StMartin’s Press, Inc. New York.
Renfro, R.Z.H., 1992.” Fertilizer Price and Subsidy Policies In Bangladesh.”

World Development, Vol. 20, No.3
Sadoulet, E., De Janvry, A., 1995. Quantitative Development Policy Analysis.

The Johns Hopkins University Press, Baltimore And London.

Salinger, B.Lynn., 1991. ” Exchange Rate Policy And Implications For Agricrdtural
Market Integration In West Africa.” Paper Prepared for the U.S. Agency for
International Development under Subcontrat to the Harvard Institute for
International Development.

SandersJ.H et al, 1996. The Economics of Agricultural Technology in
Sendarid Sub-Saharan Africa. The Johns Hopkins University Press,
Baltimore and London.

Sijm. J, 1992. Food Security and Policy Interactions in Mali.

106
Tinbergen Institute, Erasmus University, Rotterdam.

Staatz, M.J., 1994. ” Use of Simple Price Analyses for Analyzing the Impact
of the Cfa Franc Devaluation. ” Institut Du Sahel,
Michigan State University.

Staatz, M.J., 1989. ” The Role of Market Conditions in Influencing the
Adoption of New Agricultural Technologies in Mali.”

Report to USAID/Mali Prepared under Food
Security in Africa Cooperative Agreement Department of
Agricultural Economics. Michigan State University.

Staatz, MJ, et al., 1994.” Monitoring and Analysis of the Effects of the
Devaluation ofthe Cfa Franc on Food Security and Economic Growth
in West Africa.” Agency for International Development.

Michigan State University.

Stefl'en, N.P., 1995. The role and Limits of the Grain Market In Assuring
Household Food Security In NorthEastern Mali: Implications
For Policy.Dissertation for the Degree of ph.D
Michigan State University.

Stoorvogel, JJ. and E.M.A. Smaling, 1990. ” Assessment of Soil Nutrient Depletion in

Sub-Saharan Africa, 1983-2000.”Wageningen,

The Netherlands, DLO Winand Staring Centre for Integrated Land,
Soil and Water Research (SC-DLO), Report 28.

Stovall, J.G., “ Sources of Error in Agrregate Supply Estimates.”

107

Journal of Farm Economies, 1966.
Streeten, P., 1987. What Price Food Agricultural Price Policies in
Developing Countries. St. Martin’s Press, New York.
Stryker, J.D., Dethier, J.J., Peprah, L, Breen, D., Ievin, E.L., Wohl,
J.B., 1987.”Incentive System And Economic Policy Reform In Mali.”
Associate For International Resources And Development, Inc.287
Alewife Brook Parkway, Somerville, MA. 02144.
Tefl’t, J.F., Weber, TM and Staatz, M.J., 1990. Research Methods in the
MSU Food Security In Africa Project.
Conceptualizing and mplementingPolicy Relevant Studies.

Tessio, K.E.M., 1996. 5’ Regional Variation in Efficiency of Fertilizer Use:
The Case of Food Crops Producers in Togo.” Economies of
Agro—Chemicals Symposium, Wageningen Agricultural University,
the Netherlands.

Traore, B., 1979. Introduction of New Technology and Optimal Farm Plans.
A Linear Programming Analysis of the Household Economy in the
Subsector of Sirakorola, O.A.C.V., Zone Mali

Upton, M., and Dixon, J.M, 1994. Methods of Micro-level Analysis for
Agricultural Programmes and Policies. A Guideline for Policy Analysts
Food And Agriculture Organization of the United Nations. Rome, 1994.

Weijenberg, S.,1993. “ Rapport de la Mission d’Appui au Fond De

Développement Villageois”, Office Du Niger, Mali.

 

108
Yang, W.Y, 1965. “ Method Management Investigations.” Agricultural
Development Paper No. 80, Rome , FAO.
Yaron, D. .“ Empirical Analysis of the Demand for Water by Israeli

Agriculture” Journal of Farm Economics, 1967.

a Okra [Lav £4.51} x/i Jd

4
-
a.
......
.....
.
o
' o
. --
a
o
n
c
......
c
c
v
.-
c
c
......
.

iCHIGAN STATE UNIVERSITY LlBRARlES

W21 W |
M \h‘ WW.“ um

1293 02563 4027

q”