‘3...
a: .i . ..

.35.; y
57-} {I

 

)3

Ermfimfi ., u . . . ,, 4 . .4 ., Jrgaghwu .. . . .fiufiflwmia a

 

THESIS

Q
300‘

 

LIBRAP" I
“michlgan a!
University I

_.. w “'

 

This is to certify that the

thesis entitled

GENDER-DIFFERENTIATED HOUSEHOLD
RESOURCE ALLOCATION -
EMPIRICAL EVIDENCE IN SENEGAL

presented by

Janet M. Owens

has been accepted towards fulfillment
of the requirements for

M. 5. degree in Agricultural Economics

 

   

 

Major professor

Date fl? 71.7?00/

0-7 639 MS U is an Affirmative Action/Equal Opportunity Institution

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIFIC/DateDuepes-nts

 

 

 

GENDER-DIF F ERENTIATED HOUSEHOLD RESOURCE ALLOCATION -
EMPIRICAL EVIDENCE IN SENEGAL

By

Janet M. Owens

AN ABSTRACT OF A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Agricultural Economics

2001

Professor John A. Strauss

ABSTRACT
GENDER-DIFFERENTIATED HOUSEHOLD RESOURCE ALLOCATION -
EMPIRICAL EVIDENCE IN SENEGAL
By

Janet M. Owens

This paper examines the weak condition of pareto efficiency maintained by the unitary and
collective models of the household. Farm production data collected from Senegalese
households over a two-year period are used to test whether resources are allocated efficiently
across male- and female-managed plots within a household. Coefficient estimates show that
gender-based discrepancies in input usage across plots ultimately lead to lower yields on
female-managed plots. Across the two years, female plot managers, on average, generated
between 18% and 35% less revenue per hectare than did male plot managers living in the
same household. These results suggest that Senegalese households do not achieve allocative
efficiency in farm production because resource decisions within the household are driven at

least partially on the basis of gender.

To my Mother

who wasn’t able to see this come to fruition

iii

Acknowledgments

I extend my deepest gratitude to my thesis advisor, Professor John Strauss. His
intellectual rigor and unfailing support had a profound impact on the research and on me,
personally. I would also like to acknowledge the significant contribution of my other
committee members, Proféssor Les Manderscheid, who served as an advisor throughout my
studies, and Professor Torn Reardon, who shared his valued insights on the research issues.

Without the support of Valerie Kelly and Bocar Diagana this research would not have
been possible. Valerie provided access to the IFPRI/ISRA data set and both Bocar and
Valerie provided needed consultations to help me understand the enigmas of the data.

During the course of my graduate studies I was fortunate to develop many
relationships that not only enriched my life while at MSU but will accompany me into the
future. I thank Nazmul Chaudhury, Jim Myers, Dave Dibley, Kei Kajisa, Jing-Yi Lai, and
Takashi Yamano for their friendships, their shared intellectual pursuits, and for their levity.
They helped me find joy in everyday moments and maintain sanity while under duress. I
especially acknowledge the friendship of Nazmul Chaudhury whose exuberant spirit
continues to make me smile and remind me to live life largely. I am indebted to Mark Haas
at the Michigan Department of Treasury who contributed personally and professionally to
my intellectual development and to Larry Hembroff at the Institute for Public Policy and
Social Research who helped foment my interest in survey research. I thank Georges Fauriol

for his many years of devotion, counsel, and confidence in me. He ultimately pushed me to

iv

pursue graduate studies.

I owe special thanks to my family. They seemed to understand the moments when
I most needed their support. I thank my father for his unwavering support and generosity.
I will always strive to emulate his moral character. Finally, I thank my mother. During the
happy moments of my life, I will cherish her buoyant spirit; during the sorrowful ones, I will

grasp for her courage.

 

Table of Contents

List of Tables .............................................. vii
List of Figures .............................................. ix
Introduction ................................................ 1
Setting .................................................... 4
The Household Model ........................................ 6
Empirical Estimation ......................................... 11
Data .................................................... 12
Results .................................................... 15
Implications ................................................ 60
Conclusions ................................................ 63
Appendices ................................................. 64
Appendix A .......................................... 64
Appendix B .......................................... 67
Appendix C ............................ I .............. 74
Appendix D .......................................... 81
References ................................................. 88

vi

 

List of Tables

Table 1
Mean Revenue, Area, and Input by Gender of Field Manager .............. 16

Table 2a
1989 Plot-level Revenue per Hectare with Household and Crop Effects ...... 21

Table 2b
1990 Plot-level Revenue per Hectare with Household and Crop Effects ...... 22

Table 3a
OLS Fixed Effects of Plot-Yield Relationship1989 — All Crops
(Household and Crop Effects) ....................................... 24

Table 3b
OLS Fixed Effects of Plot-Yield Relationship1990 —- All Crops
(Household and Crop Effects) ....................................... 25

Table 4a
OLS Fixed Effects of Labor Input Intensities 1989 — All Fields
(Household and Crop Effects) ....................................... 28

Table 4b
OLS Fixed Effects of Labor Input Intensities 1989 — Peanut Fields
(Household and Crop Effects) ....................................... 30

Table 4c
OLS Fixed Effects of Labor Input Intensities 1989 — Millet and Sorghum Fields
(Household and Crop Effects) ....................................... 32

Table 4d
OLS Fixed Effects of Non-Labor Input Intensities 1989 — All Fields
(Household and Crop Effects) ....................................... 33

Table 4e

OLS Fixed Effects of Non-Labor Input Intensities 1989 — Peanut Fields
(Household and Crop Effects) ....................................... 35

vii

 

Table 4f
OLS Fixed Effects of Non-Labor Input Intensities 1989 —- Millet and Sorghum Fields
(Household and Crop Effects) ....................................... 36

Table 5a
OLS Fixed Effects of Labor Input Intensities 1990 — All Fields
(Household and Crop Effects) ....................................... 38

Table 5b
OLS Fixed Effects of Labor Input Intensities 1990 - Peanut Fields
(Household and Crop Effects) ....................................... 39

Table 5c
OLS Fixed Effects of Labor Input Intensities 1990 — Millet and Sorghum Fields
(Household and Crop Effects) ....................................... 41

Table 5d
OLS Fixed Effects of Non-Labor Input Intensities 1990 — All Fields
(Household and Crop Effects) ....................................... 42

Table 5e
OLS Fixed Effects of Non-Labor Input Intensities 1990 — Peanut Fields
(Household and Crop Effects) ....................................... 43

Table 5f
OLS Fixed Effects of Non-Labor Input Intensities 1990 - Millet and Sorghum Fields
(Household and Crop Effects) ....................................... 44

Table 6
Household Dispersion in Yields ..................................... 53

viii

 

List of Figures

Figure 1
Production Residuals, 1989 ......................................... 57
Figure 2
Production Residuals, 1990 ......................................... 58

ix

 

INTRODUCTION

The family plays a pivotal role in the allocation and distribution of resources . Yet
much controversy exists over how these processes occur. The early analysis treated the
household as an aggregate ignoring the specifics of member outcomes and preferences.
Becker’s (1965) theory of time conceptualized households as allocating time between the
production of goods at home and the purchase of goods with income earned from
participation in market activities. Recognizing that agricultural production generates a
primary source of income for rural households in developing countries, Singh, Squire,
Strauss (1986) extended the analysis of agricultural household behavior by acknowledging
that small farmers participate as both producers and consumers in commodity markets to
varying degrees'. Individuals buy and sell inputs and commodities in the market while
consuming some of their production and utilizing their own resources, namely labor, for

production. Chiappori (1988 and 1992) among others2 recognized that the unitary model of

 

The degree depends on whether farmers are net buyers or net sellers. Using data from
Senegal, Goetz, 1990 and 1992, models this decision as a two-stage process: first, do
farmers participate in the market, and second, if so, do farmers participate as net buyers or
net sellers.

2

Manser and Brown, 1980, and McElroy and Horney, 1981 , initiated the two-person
Nash bargaining framework as an alternative to representing the household as an aggregated
entity in which members pool their resources and maximize a single utility function. These
studies offered the possibility that two individuals would consider forming a household on
the basis of whether the utility gains associated with marriage would be greater than the sum
of each individual’s indirect utility, a function of prices related to the consumption of private
goods and non-labor income. An individual’s indirect utility is representative of her best
possible alternative stateueither remaining single or choosing another partner, given

1

the household was neither illuminating about the real processes of distribution occurring
within the household nor fundamentally linked to neoclassical theory. This work
demonstrated that households should be represented as a set of members who may have
distinct goals and preferences. Using consumption data, Browning, et a1 (1996), Thomas
(1990), and Hopkins and Haddad (1994), showed that either different types of expenditures
made in the household or the timing of these expenditures could be linked to the shares of
income claimed by the various members. While these studies disputed the relevance of the
unitary model, they made only minimal assumptions about how members resolve conflicts:
The outcome of the distributive process was generally posited as being an efficient outcome.

Confronting the weak condition of pareto efficiency, Udry (1996) tested whether the
behavior of agricultural producers in Burkina Faso could be assumed to meet this criteria.3
His findings showed that households allocated agricultural inputs to members in such a way
as to cause losses in agricultural output. His results suggest that neither the unitary nor
cooperative model suffice to explain household behaviOr.

Following the important contribution of Udry’s (1996) inquiry into intra-household

 

information about possible alternatives. The economic gains associated with the marriage are
evaluated according to the indirect utilities of the individuals which become threat points to
the potential dissolution of the union.

Lundberg and Pollack, 1993, tempered the notion of the fall-back position by
suggesting that a non-cooperative marriage could result in the partners retreating to their
‘separate spheres’ and not necessarily in divorce.

3

Jones, 1983, tested whether labor supply behavior of women living in agricultural
households in Cameroon could be characterized as allocatively efficient. She concluded that
compared to widows, who maintained complete control over the remuneration of their labor,
wives spent 40% less time in rice planting activities that generated higher returns to labor
than other crops but which were expected to be handed over to men. Under circumstances
in which the returns to women’s labor are contested by male household members who hold
different preferences, Jones suggested that the problem associated with dividing the proceeds
will be resolved by bargaining.

resource allocation in Burkina Faso, this paper will explore whether similar findings can be
supported in Senegal, another west African setting. Operating within the neoclassical
fiamework of separability in supply and consumption decisions, and on the weak assumption
maintained by the unitary and collective models-- namely that allocations of household
resources are efficient-- I test whether the gender of the plot manager affects input usage
across fields within a household whose members manage separate fields using two years of
cross-section data. Similar to Udry, I control for the heterogeneity of resource endowments
across households by using an OLS dummy variable regression approach. The results
obtained from this method are analogous to those generated by using a time-demeaned fixed
effects approach with panel data. This strategy enables me to isolate the comparison of input
usage across male- and female-managed fields planted to the same or similar crops within
a particular household and without having to match fields over time. The matching of plots
is not possible in this setting because land was both reconfigured and reassigned to different
household members for each of the agricultural seasons occurring during the survey period.
I examine the gender-specific effects separately across the two-year period because
agricultural conditions appeared to be extremely time-sensitive, although pooling the two
years of data may have led to more robust results. I find gender-based discrepancies in input
usage across plots within a household which ultimately lead to lower yields on female-
managed plots. These findings provide further corroboration that neither the unitary nor the
cooperative bargaining models used currently to analyze the determinants of intra-

household decision making are realistic.

SETTING

Most ethnic groups in Senegal live in extended family units and maintain polygamous
households.4 The extended family may be comprised of more than one nuclear family,
unmarried siblings, and other dependents. These families reside in a vertical hierarchy of
succeeding generations, known as a concession, in which married sons and their dependents
are subordinate to their father. It is the father, or head of concession, who determines
agricultural production by delegating available land as communal (family) fields or personal
fields.5 As an obligation to the household head, all members must allocate some of their
labor to the household fields which are utilized for the production of food crops. In
exchange, the household head provides food for the family and land for personal use.
Women are entitled to land for their own use which they may allocate to cash crop
production. Some of the returns generated from women’s personal fields are used to provide

condiments, including vegetables and dried fish, to the meal. In addition to working in

 

About 43% of the married male household head respondents had more than one spouse.
While this occurrence is higher than the 31.3% reported by Goldman and Pebley, 1989, who
use 1976 census data, it may be comparable. Census data report the incidence of poly gyny
for all man'ied men, whereas the IF PRI/ISRA data contain a large number of concession
heads who would be wealthier on average than a nationally representative household sample.

5

Men and women farm separately throughout many societies in West Afiica, but the
allocation of land by men to women and their ultimate control of the agricultural production
process has historical precedence. Guyer (1980) defines the sexual division of labor for the
Beti, an ethnic group in Cameroon: During precolonial times men’ 5 contributions to farming
was substantial. They cleared the forest for planting, felled trees, built enclosures to protect
crops from destruction by animals, and constructed storage houses for crops. These fields
were considered owned by men although women weeded, harvested and took general care
of them. However, it is the men’s work which defined field size and length of fallow.

In modern times, by making annual plot assignments to household members, men still
determine the field size and length of fallow for land utilized as personal plots within the
family.

family and personal fields, women perform other home maintenance activities and may be
engaged in off-farm income activities.

Thus, the pattern of agricultural production in Senegal corresponds with the pattern
found in Burkina Faso. Household production is implemented on multiple parcels of land and
controlled by different members of the household. Crop choice and input decisions are made
independently by each of the members who are allocated land while the household serves as
a pool of labor for production. Men, women, and children perform specific agricultural tasks
determined usually on the basis of gender. In the Senegal context, agriculture is managed
within the household by multiple individuals but is dependent upon the complementarities
of male and female labor supply. Additionally, family plot managers who grow peanuts, the
family’s most important cash crop, may be dependent upon the household head for obtaining
peanut seed. Peanut seed has been cited by Senegalese farmers as one of the most critical
input constraints (Kelly, 1 996).

The dynamics of interdependent household production in Senegal generate a setting
in which plot managers are successful only if they can negotiate for shares of various inputs
from other household members. When members fail to obtain the inputs necessary for
optimal productivity they may incur losses attributed to inefficiency. Particularly when credit
is scarce and the only available resources obtained by the household head are not dispersed
to other members, the only asset under a women’s domain is her productive labor women.
Goetz (1990) suggests that labor cooperation between men and women may fall apart during
periods of low food productivity. He depicts the Senegalese household as a coalition of
individuals with divergent interests. The household head is charged with the responsibility

of growing a sufficient amount of cereal to meet nuclear household food requirements.

Wives, young sons, and migrant workers, in contrast, are focused on generating cash income.
Each of these members in the household pursues a distinct objective, but economies of scale
will be obtained if these members form a coalition. If the sum of the benefits obtained by
the individual members working together as a group is greater than the sum of the individual
member’s gain from working alone, cooperation should be sustainable. However, if the

benefits received by the coalition differ from those contracted, the coalition may fail.

THE HOUSEHOLD MODEL

Pareto efficiency in a cooperative agricultural household implies that factors be
allocated efficiently across its productive activities. Consider a household with two
members, a male, m, and a female, f, who produce a crop, (qc), on separate parcels of land
owned by the household.“ The crop is produced with two inputs of labor (Lim) and (L,’) on
each of two land parcels embodying characteristics of size and quality. The household has
an endowment of land, comprised of multiple parcels, (A, ), and labor time, (LT).7 Although
these members apply their labor to the production on both land parcels, they allocate labor
to other activities associated with home production, 2, and to off-farm production, qo. Since
the household’s objective is to maximize the profits achieved fiom its fixed assets of land,
it will utilize labor until the marginal revenue product of labor is equated to a shadow

wage—or market wage depending on the existence of a well functioning labor market. In a

 

The model can be generalized to a multiple person, multiple crop, and multiple plot
setting.
7

Although the data Show that some of the sampled households employed hired labor,

most of households didn’t seem to rely upon the hired labor market. Therefore, hired labor
is not considered in the model.

recursive setting, where market substitutes may be obtained for family labor or home-
produced goods, the household will determine labor demand independently of its tastes and
household composition.8 Considerations such as the gender of the plot manager or the
bargaining weight which determines a particular member’s income share shouldn’t be
factored into production decisions. Thus, male and female household members allocate
their time to crop production managed on their own field and on” the field of the other
member, off-farm work, and home production.9 The production of crop qc is a function of
G,(L,-"', Li’, Ai ), where G i(.) embodies the technology associated with a particular crop. The
optimal technology choice available to a household is conditioned by its ability to access
factor, commodity, and credit markets. These conditioning factors will be time-varying.lo

Technology, defined by the level, intensity, and timing of input usage on a particular
plot, plays a pivotal role in the assessment of whether households allocate productive
resources efficiently. Although we would expect optimal technology choice to vary by crop,
we would not expect optimal choice—or having access to the choice-to be sensitive to
household parameters such as the gender of the plot manager. In this scenario, both

allocative efficiency and the separability of agricultural factor demand decisions from

 

See Singh, Squire, Strauss, 1996, and Benjamin, 1992, for a full treatment of
separability and recursivity in agriculture household models.

9

It is possible that time allocated to one or more of these activities is zero for one of
the members.

I0

de Janvry, et a1 (1991) suggests that market failures are generally not pandemic,
rather markets fail selectively for a particular household. Thus, a market failure occurs when
the gain from utilizing a market is below associated costs.

7

household supply characteristics would fail.ll

Consumption is comprised of goods produced on farm and goods purchased with
proceeds generated by sales of tradeable crops or by off-farm income activities. Goods may
be purchased for either private consumption or public consumption. Privately consumed
goods may be defined as goods assignable to a particular individual whereas publicly
consumed goods are not separable across individuals. Leisure, If and 1'“, is separable
between the two individuals. Output prices are normalized to one.
Thus, the household’s problem is to

maxUm(cm,cf,z,lm,lf)+Auf(cm,cf,z,lm,If)

subject to :

Y = G (L,"‘, L,‘, A,) + G(L2"‘, LZ’, A2,)

Z = 2 (2f , z'“)

T f: (L,f + L2f+ qof+ zf+ If)

m =(L,m +L2m+qom+zm+1m)
P... C... + RC. s [(Yo - w... (L.'" + L?) - w. (Li + L25) + w... q."‘ + quo‘]

A=A,+A2

 

Benjamin (1992) conveys an obvious example of labor demand and supply
separation: “. . .with separation, the number of workers in Baron Rothchild’ s vineyards should .
not depend on the number of daughters he has.”

The separability condition may fail for a household when one or several factor
markets are either nonexistent or malfunctioning. Previous studies using data from Afi'ica
have documented the impact of missing labor markets on peasant household labor allocation
(Barrett, 1996; Collier, 1983; Udry, 1998). Responding to labor market failures, farmers--
differentiated by their existing endowments of capital and labor--will utilize their labor
according to specific household supply parameters, or a household shadow wage. Thus,
these disparities in factor endowments result in labor marginal productivities that are widely
dispersed across farms. Differentiated labor use intensity across farms of different size has
been associated with the inverse farm size productivity debate.

8

The maximized value of U(.) is increasing in income. So, the problem can be solved by first
maximizing income, or production with respect to labor, land, and technology, and then
maximizing utility.
The household’s production problem is to
max [G(L,"‘, L,’, A,) + G(L2"‘, LZ’, A2.)] - wlm (le +L2'") - wf (L,f +L2’)
s. t. A, + A2 = A
This generates 4 productive efficiency conditions:

(1) a Gm,m.L,‘.A,_)- wm =0
am

(2) a GQImLLIfs AI.) " WI = O
a L,‘

(3) BOILELILAZJ- w... =0
a L;

(4) anguzng- w. =0
a L;

(5) (19 = 50

6A, a 2
We can equate conditions (1 ) with (3) and conditions (2) with (4) to get the following
two conditions:

(1) a—GLLIEIEAlJ = (3) aflflzfidfi‘z.)
(9le a Lz'“

(2) afifLiZLlLA. .l = (4) aflflLLAzJ
The marginal product of men’s labor allocated to their own plots Should be
e(Inivalent to what men allocate on women’s plots, the same for women’s allocation of

labor. Solving the system of 5 equations and 5 unknowns from the first order conditions

9

we get the endogenous factors of labor (le, Lz'“, L,’, L}, A,, A2 ) as functions of (w, wm,
A).

Household members growing the same crop on similar plots should produce
equivalent yields. This condition of constant returns to scale should hold if plots managed
by household members were of the same size, comprised Similar levels and qualities of
micro- and macro-nutrients, and were exposed to similar agro-climatic conditions.

Households choose labor and land to maximize income. Thus, optimal production
decisions depend only upon input prices and plot characteristics and they should be
independent of household parameters. In other words, the choices made towards the
production of the same crop occurring on separate plots, controlling for soil characteristics,
within a particular household should be similar, irrespective of the gender of the plot
manager or the preferences expressed by these two individuals.

The process by which resources are allocated to the preferences of these two
individuals is another matter. In a cooperative setting, allocation might be considered as a
two stage process.12 First, income would be allocated towards public goods consumption,
or to those items that are not identified uniquely with a particular member, and to each of the
members for expenditure on their own consumption. The issue of jointness may be
particularly complex when considering the polygamous household. The consumption of

food, for example, would be difficult to identify with a particular member, but it is possible

that it could be assignable to a wife and her children, a sub-unit of the extended household.

¥

I2

Browning,et al, 1996; Chiappori, 1988 and Chiappori, 1992 suggest the two stage income
allocation rule as a plausible candidate for the intra-household allocation process. Chiappori
(1997) generalizes the resource allocation process as a function of both members’ wages and
noIl-wage incomes.

10

Second, each member spends her or his portion of total household income on nonpublic
goods. The allocation of income across members would be based on multiple factors,
including shadow wages or economic opportunities available outside of the household and
predetermined agreements for income sharing concluded as part of the household formation
process. However, the sharing rule should not influence household income generation
decisions or member preferences.
EMPIRICAL ESTIMATION

The state of productive efficiency is determined exclusively by technology and the
inherent characteristics of the inputs and fixed factors used to grow crops. When two plots
of land are used to grow the same crop in a household, the only differentiation in outcome
should result from differences occurring in soil characteristics between the plots. For the
same crop, technology choice, which embodies the timing, intensity, and type of labor inputs,
should be identical. ‘3 Therefore, the empirical content of the paper examines whether the
deviation of plot yield from the mean yield of a household is related to the gender or status
of its plot manager.l4 A general output supply equation is estimated to account for
differences in technology use resulting from crop choice and for factors that condition a

household’s ability to engage in market activity:

(6) thi=BO+ 91 thi + 92 thi + I33 Vh + B4 Yc+ €th ,

 

Land quality may affect technology choice decisions. Therefore, if two household
members were assigned plots of different quality, technology choice could vary over these
two parcels of land.

I4

Productive efficiency should also incorporate decisions of land allocation: Land should
not be allocated on the basis of the gender of the plot manager. Rather, the household head
should allocate land based on the marginal revenue product of the crops grown on each of the
household’s land parcels.

11

where th, is output/hectare obtained on a given plot; X he, is a dummy variable capturing
plot-size effects; Ghci is the gender of the individual who controls the plot; V b is a household
fixed-effect dummy that restricts attention to the variation in yields across plots planted
within a single household; y, is a crop dummy that controls for the impact of technology-
specific crop effects; and Em is an error term that summarizes the effects of unobserved plot
quality variation and plot-specific production shocks on yields. Similarly, an input demand
equation is estimated for each type of household labor input and for selected non-labor inputs
to examine whether differences in input intensities across fields within the same household
may be attributed to the gender of the plot manager. If gender influences the underlying
household decision rule in factor allocations, then [32 would be significantly different from
zero in these estimations. The results are estimated separately for the two years.
DATA

The data used for this paper were collected under the International Food Policy
Research Institute (IFPRI) and the Institut Senegalais de Recherches Agricoles (ISRA) study,
Supply and Consumption Impacts of Agricultural Price Policies in the Peanut Basin and
Senegal Oriental. The survey, covering approximately 300 households, comprised three
years" of household panel-data collection and was designed to learn how changes in
agricultural policy affect household behavior. Coverage was focused on the Peanut Basin,
an area comprising one-third of the country’s land area and over two-thirds of its rural
population. The data provide detailed information on rural and urban household

consumption and production patterns, including both farm and off-farm activities.

 

15

The survey commenced during the 1988 harvest so detailed crop information are
available for only two complete farm-year cycles.

12

Enumerators obtained data on labor and non-labor input usage on each plot throughout the
farm cycle for two seasons. These efforts generated usable data on approximately 2700
plots.16 Data collection comprised 18 separate surveys on the following topics: household
demographics; food consumption; purchases and sales related to crops and livestock;
expenditures on all other services and products; cash transfers; household assets; individual
net income from all economic activities; gross income and input costs; and detailed
production data by plot for 1989/90 and 1990/91 crop seasons. Sample selection was based
on an earlier reconnaissance survey that synthesized information on general characteristics
of the area.
Data Issues

Plot-level Revenue This variable is the product of total output harvested per hectare
on a plot and the average village1989 commodity price. These prices were used for both
cross-section years to facilitate the comparability of results. A number of fields were planted
with multiple crops which are not comparable to mono-cropped fields.17 Thus, these fields
were excluded from the crop-specific analysis but were incorporated into the all fields
analysis.

Labor Inputs Detailed labor information is available from the survey on the hours

 

16

During the first year of the survey each plot was measured, but in the subsequent year
only a subset of plots was measured directly or estimated.

I7

Benjamin (1995) speculates that studies using a measure of aggregate value output
to estimate the relationship between farm size and productivity could have led to a bias
tOWard finding the frequently observed inverse relationship. The bias would be particularly '
sensitive when aggregate output comprises both high- and low-value crops and data on land
qualiry are not available. If high quality land is more expensive-which is observed by the
PYOducer- then efficiency would prescribe that the more expensive crops be planted on it.

13

allocated to each field by activity and for each type of labor input: household male, female,
child, and owned animal equipment hours as well as for various types of hired or in-kind
labor. Because hired labor is employed only sporadically by these households, I do not
estimate the effects of hired labor usage on yield outcomes.

Non-labor Inputs Although field applications of all inputs are detailed in the survey,
usage of any particular input is not widespread. Accordingly, I estimate usage for only seed,
manure, chemical fertilizer, and fungicide. With the exception of seed, I translated the non-
labor input quantities into dummy variables and estimated their usage as a linear probability
model because they were measured imprecisely.

Commodities In addition to estimating the models on all of the relevant fields for
each year, I selected millet, sorghum, and peanuts as representative crOps grown by men and
women and provide separate results for fields cultivated with either millet or sorghum and
for fields cultivated with peanut. These choices are useful because they typify a food staple
grown for household consumption and managed by a household head, and a cash crop
intended for market sale by both males and females. Peanut production accounts for the
largest share of cash income from crop production, although some of the harvest could be
reserved for consumption. Millet and sorghum could also be planted as cash crops, but it
is more likely that they would be produced for home consumption. I combine the millet and
sorghum fields to achieve more robust results. It is not possible to identify explicitly fields
as either communal or personal. However, it is likely that a field planted with millet or
sorghum and managed by a household head will be considered communal.

Plot Size I transformed continuous plot size data into a categorical variable of plot-

size quintiles. The grouped data elucidate the plot-yield relationship more clearly.

14

RESULTS

Table 1 displays descriptive statistics of agricultural yields (value per hectare) and
input intensities per hectare for the two survey years by gender of plot manager. In 1989,
women generated more revenue per hectare than men, although the difference between these
two groups was not significant. The reverse held true for men in 1990 and their gain over
women’s average revenues was more substantial. The relative variation of men’s and
women’s revenues between the two years may be explained by the distribution of crops
farmed by the two groups across the two-year period. Peanut farming generates more value
per hectare than other crops and in 1990 the number of fields allocated by men to peanuts
represented 32.8% of all fields farmed by both men and women. In 1989, men’s peanut
fields accounted for only 25.3% of all fields. Udry reported higher average revenues for
women’s than for men’s fields in Burkina Faso, where the relative proportion of groundnuts
farmed favored women (15.6% of women’s fields compared to only 5.1% of men’s fields).

In contrast to the above mentioned variations in plot yield, the differences found in
the average levels of inputs used by men and women over both years are stark. In both years,
women farmed plots that were less than half of the area of men’s plots. Male labor and
animal traction labor were utilized more intensively on male-managed plots while females
used substantially more female labor and seeded their plots in greater densities. Men applied

more manure, fertilizer, and fungicide on their own fields, but overall use was not high.

15

Mean Revenue, Area, and Input by Gender of Field Manager

Table 1

 

 

 

 

_l989' 1.28.2 T- 1&9 M T-
Men’s Women’s Statistic Men’s Women’s Statistic
Fields Fields H,:u.=u, Fields Fields I1,,:u,,,=ur
Ag Revenue 48864.6 51189.9 -O.97 31791.8 26782.8 2.354
per Hectare (45393.1) (44387.6) (3496.6) (33302.5)
Area of Plot 1.007 .4353 18.7 .991 .436 16.2
(hectare) (.8625) (.3662) (.028) (.019)
Male Lahor’ 156.27 88.5392 8.6 113.05 77.38 6.7
(hours per (185.62) (124.01) (113.19) (68.67)
hectare)
Female Labor 26.42 173.30 -7.8 15.03 81.52 -8.2
(hours per (70.76) (416.96) (40.98) (151.83)
hectare)
Child Labor 85.00 85.15 -0.02 68.44 57.90 1.8
(hours per (166.62) (148.62) (86.98) (97.12)
hectare)
Animal Traction 49.6717 40.1506 2.9 44.17 37.16 3.1
(hours per (66.13) (57.43) (44.09) (32.37)
hectare)
Seed 39.38 63.87 -8.9 47.89 79.93 -2.8
(kg/ hectare) (63.87) (56.76) (146.93) (192.66)
Manure Use .064 .022 4.2 .07 .005 4.8
(1 if used; 0 if (.24) (.15) (.26) (.075)
not)
Fertilizer .021 .004 3 .4 .089 .081 0.46
(1 if used; 0 if (.1451) (.063) (.2849) (.2732)
not)
F ungicide .2086 .1218 4.6 .4531 .4208 2.2
(1 if used; 0 if (.4065) (.3273) (.016) (.022)
not)

 

 

Note- Standard deviations are in parentheses.
'Data was determined not to be suitable for pooling. Chow tests are provided in the appendices, tables D 1 -D3.
2 Harvest and post-harvest labor activities are not included in labor hours for any type of labor.

l6

Leaving aside differences in input intensities, other factors could explain variations
in yield outcomes. '8 Without controlling for these factors, it would be difficult to disentangle
outcomes that are linked to market phenomena occurring across all households from those
that are linked to the decision patterns concerning resource allocations to different members
within the same household. Therefore, it’s important to examine yield variations between
men and women farming the same crop within the same household during the same year.

Tables 2a and 2b provide evidence that plots controlled by women produce
substantially lower yields. The All Crops columns in both tables report estimates of equation
(6), yield differences for all of men’s and women’s fields while controlling for household
and crop effects. Referring to the Column (2) Specification under All Crops, women
generated 35% less revenue per hectare from their plots than did men on average in 1989.19
This striking difference in disparity of outcomes by gender was repeated in 1990. Women
generated 27% less then the average yield per hectare found on men’ 5 plots. These estimates
provide further justification to Udry’s claim that productive household resources are not

being allocated across members in a pareto-efficient manner”.

 

Udry provides several explanations contributing to differences in men’s and women’ S
yields within the same household. One, systematic variations in nutrient soil quality across
men’s and women’s plots would exacerbate differences in yields by gender. Two, customary
crop choices by gender would generate different yield outcomes. Three, the prevalence of
inefficient land and labor markets would create distortions across households in their
capacity to use factors of production efficiently resulting in different factor shadow prices.
Four, nonexistent credit markets could distort factor shadow prices across time. All of these
factors could be supported within the Senegal context.

19

The gender differential in yield is computed as the percentage difference from
average household yield.

20

Udry found that women’s yields were reduced by 30% of the average yield on plots
farmed in Burkina F aso.

17

The Peanut and Millet/ Sorghum columns restrict estimates of gender-differentiated
yields to fields cultivated exclusively with peanut, millet or sorghum. Differences in yield
outcomes between men’ s and women’s fields remain substantial. Female peanut cultivators
produced 24.4% less output per hectare than the average peanut yield in 1989 and 19.7% less
in 1990. Yield differentials between male and female cultivators of millet and sorghum
fields provide even more stark contrasts. Females growing millet and sorghum produced
52.7% less output per hectare on average than their male kin. In 1990, this production loss
climbed to 63.5%.

The gender effects on yields reported above control for plot size - yield effects.” All
of the 1989 specifications incorporating plot-size quintiles and the 1990 all crops
specification demonstrate a strong negative relationship between plot size and yield.
However, unobserved differences in input intensity, plot characteristics, or prices may be
correlated with plot size and/or the assignment of plots by the household head to other
members. If any one or combination of these factors underlies the inverse plot size - yield
relationship, then plot size becomes endogenous in the above specifications.

For example, cultivators of smaller parcels of land may use inputs more intensively.
Most probably, smaller parcels of land would be identified as personal plots compared to the
larger ones identified as family, or communal plots. Input differences across personal and
communal plots may ascribe to differences in the quality and timing of labor inputs.

Individuals would be inclined to allocate more efficient units of labor to their own plots

 

21

The frequently observed inverse relationship between size and productivity in the
literature refers to farm size and not plot size.

18

compared to the labor they expend on communal plot production. Therefore, communal plot
managers would have to allocate time to the monitoring and supervision of family labor to
obtain equivalent labor inputs. Monitoring and supervision time represent additional costs
to the communal plot manager.

Household heads may respond to variations in land quality by dividing land of higher
quality into smaller parcels, or they may own noncontiguous parcels throughout the village
that were sized according to the underlying characteristics. If so, then unobserved soil
quality would be inversely correlated with plot size and the household head may be
assigning plots nonrandomly on the basis of soil quality. In this scenario, the gender and
plot-size coefficients would be biased downward when soil quality is omitted from the
equation.

In contrast to the above notion that small plots embody superior soil characteristics,
plot size may also be correlated with distance from the household. The afore-mentioned
relationship may describe plots that are located only within some concentric interior.
Although no information about the distance of plots from the household exists in this data
set, previous Senegalese studies depict land belonging to a particular household as being
organized within concentric circles around the compound.22 Women are allocated parcels
of land that are located towards the periphery of the household’s holdings. Not only would

the large distance imposed on the managers of these plots represent an additional cost, but

22

See John Waterbury , “The Senegalese Peasant: How Good is our Conventional
Wisdom?” in Gersovitz and Waterbury, 1987, for more details on rural livelihoods and the
c_°13ir1g mechanisms employed by Senegalese households to confront prevalent conditions of
risk. Waterbury notes that women’ 5 personal fields, along with those provided to non-family
labor, are located the farthest away from the household and considered ‘bush fields’. The
reference suggests fields of marginal quality.

19

these plots would be of lesser quality. Women would be farming plots that are not only
smaller than the average Size of a household plot, but are of inferior quality compared to the
average parcel of land owned by a household. Thus, when I control for plot size I may be
comparing yields on male- and female- managed plots that exhibit substantial differentials
in quality. This comparison would inflate the gender-differentiated effect in yield outcomes.

Without controlling for crop or household effects, Table 1 showed that women farm

on substantially smaller plots than men. If it is the case that women are allocated smaller
plots on a systematic basis and smaller plots produce higher yields, we would expect the
impact of gender on yield variation to decrease when controlling for plot size. The Column
(1) specifications in Table 2a and Table 2b report estimates of gender yield differentials
without plot-size quintile dummies but control for crop and household fixed effects
estimates for all fields and household fixed effects for the specific crop estimates.
Comparing these column estimates to those that control for plot size, I find that the gender
effect is smaller when the inverse plot size relationship holds true. In 1990, the inverse plot
size - yield relationship did not hold for peanut fields.

The gender coefficient is smaller in each of the Column (1) specifications compared
to the gender coefficient reported in the corresponding Column (2) specification because it
is picking up the average differential in yields associated with the inverse Size effect.
Though smaller, the Column (1) gender coefficient identifies the total gender effect, whereas
the Column (2) coefficient a partial gender effect. On average, women are allocated smaller
parcels of land--which is further substantiated by the reduced form estimates of input
intensities discussed below--and the total gender effect captures this yield effect.

I attempt to disaggregate the average effect of plot size on gender by estimating the

20

38.85 E 0.8 32.3.; ”$35.83.. 5 8:33..-. we o=_m> 8293< No.02

 

 

 

mm: mm: a: R: 3m. 2:. 222... .o 52.32
3. mm. 3.. a. 3.. 2. N2
.8... SN 8...... :2 8...... 2.: 8.55.5 a...
8...... gm .8... m2 8...... a... 8228.. .8... 8.. 8...... 8.. 3.5.55 E.
mutmtfim Am
3.5%.” 2.8m: 383 228%..532
w< Co :82
G... .V .8... and .8. .v G . .9 3......
8.32 a... .2: 3.38.: 2.3.3 ~23? an. . 28 .5238
83.8 and :3.
8.3.2- 3.. .48- a. . 82. 2:23 3
2...... .33 5.8
3.32:. 8.33. 3.842- 2:23 3
$2. 828 GS.
SASS- 3.22.- 8. 8.....- 2:23 3
33... 3...... ES
:22... . - ”$88- 2.3 .2- 2:23 2N
Nona SE
8m... 6...: 5.0 5.9 :3. 5.3
8.22.- 3.33. a... .22. 3.3%.- 8.3:..- .383- .330
a. S E 2% 5 €
EzstmboE—z Samoa 39.0 =<

 

83b”..— aeLU 15. 22.8.8: .53 9:38: 22. 2:35: .9673: $3
3 2%...

21

 

 

 

3.. 2.... 8.. on. a: . a: . 222”. .o 252:2
2. a. 8. N... an. S. .2
8...... 3.. .3322 .232... 8.22.5 a:
.3 23: 8......82 .82. .2. .oo.o..~.~. .8222 8233.: 8.22.5 :2
8:222 -h.
oodommm .2933 $6 . mom o§oo£o=§>om
w< Co :82
6.... 5.... .8... .8. .. 83. and
.23.... m $.83: 2...: .N. 2.28... 3. .3: 3.2.“? .2238
32.. cm: and
8.8: 3. .2... $.82- 2:23 3
Ga... an... .82.
3.2.2 3:3. 8.2%- 2:23 5..
.3. c 2 :8 am...
8.22 3.2.. 232. 2:23 3
GE. 5.... .2...
3.3: 8.38 8.82- 2:23 .3
”uflm SE
33.. 2.3.. 2.2. 5...... .2... 2.3.
3.23... 2.52..- :83- 8.38- E. . z.- o..: G- .2250
S S S 3 S 3
23932:: 232. 32o ..<

 

392cm 9:0 can 22.025: 5.3 9:33: com 05:96“ .9632.— 33
AN via-n.

22

yield equation with both plot size dummies and gender- plot size interactions for each
quintile, excluding the first category of each group. These results, along with those that
attempt to establish whether similar partial effects may exist across household generations,
are presented in Tables 3a and 3b. The estimates control for both crop and household effects.
The gender-plot size interactions are not jointly significant which suggests that the pattern
of yield differences by plot size does not vary according to the gender of the field manager.
Therefore, there is no empirical evidence to substantiate the claim that women sustained
lower yields on land parcels of similar size than men who planted the same crop and lived
in the same household. With the exception of one gender-plot interaction term in 1989, the
interaction terms are neither singularly significant nor jointly significant for either year. The
test statistic evaluating the strength of the gender-land relationship in 1989 is distributed as
F(4, 1364) and has a value = 1.69 (p = 0.15). The 1990 test statistic is distributed as F(4,
1008) and has a value = 0.14 (p = 0.97).

The results demonstrate that the inverse size-yield relationship do not affect male-
and female-managed plots differently. Although it is not possible to control for unobserved
differences in soil quality or location between men and women’s fields, these results suggest
that men and women farm plots of relatively similar characteristics.

If one is concerned about the nonexogenous nature of plot size allocation and gender,
Stature within the household may suggest a potential source of additional endogeneity:
Assuming that women are allocated the smallest plots, it is plausible that household heads
1‘ eServe the largest and most productive plots for themselves. Therefore, column (2) in these

tW0 tables estimates the impact of headship status on yield to identify whether systematic

biases occur between the head and other male members. These estimates were based only on

23

Table 3a

OLS Fixed Effects of Plot-Yield Relationship
1989-A11 Crops (Household and Crop Effects)

 

 

(l) (2)
Gender-Plot Size HH Status- Plot Size Effects
Interactions Only Male—Controlled Fields
All Fields Ag Revenue
Ag Revenue (per hectare)
(per hectare)

Nonhead -14400. 140
(0= HH head, (4.54)
l=other males)

Plot size:

2“" quintile -10106.88 42785.34
(2.18) (2.73)
3rd quintile -17357.73 -l977l.98
(3.85) (4.28)
4"I quintile -25190.48 -29088.86
(5.63) (6.24)
5‘h quintile -3 1319.55 -36493.21
(6.82) (7.40)
Gender -13504.62
(2.71)
2"d * gender -l2698.36
(1.93)
3" "‘ gender -2769.70
(0.41)
4‘h * gender 5274.65
(0.59)
50‘ "' gender -2518.48
(0.22)
Constant 56359.99 73744.13
(2.09) (2-75)
F- Statistics
HH Dummies 1.87 [0.00] 1.71 [0.00]
Plot Dummies 14.19 [0.00] 15.86 [0.00]
Gender“ Plot Size 1 .69 [0.15]
R2 .39 .44
Number of Fields ISYL 11 10

 

24

Table 3b
OLS Fixed Effects of Plot-Yield Relationship
1990—All Crops @ousehold and Crop Effects)

 

 

(1) (2)
Gender-Plot Size HH Status- Plot Size Effects
Interactions Only Male-Controlled Fields
All Fields Ag Revenue
Ag Revenue (per hectare)
(per hectare)
Nonhead -8258.433
(O= HH head, (3-56)
l=other males)
Plot size:
2"" quintile -4127.54 -4446.767
(129) (1.31)
3rd quintile -401 1.37 -3467.427
(1.18) (0.94)
4th quintile -6459.73 —6608.929
(2.06) (1.96)
5th quintile -8143.01 -10744.570
(2.49) (2.93)
Gender -9860.95
(2.86)
2"‘1 * gender -2924.82
(0.65)
3rd “ gender -999.56
(0.19)
4‘h " gender -33 89.45
(0.59)
5'“ * gender 4565.88
(0.22)
Constant 38252.62 41677.19
(3.22) (3.04)
F- Statistics
HH Dummies 5.20 [0.00] 3.70 [0.00]
Plot Dummies 1.80 [0.00] 2.66 [0.03]
Gender‘Plot Size 0.14 [0.97]
R2 .58 .59
Number of Fields 1 178 845

 

25

fields managed by men. Nonhead males, like females, were found to generate less yields per
hectare than the residing head. For both years the plot quintile dummies are jointly significant
and in 1989 the inverse relationship between plot size and yield of male cultivators is strong.

The third column in each of the tables presents yield estimates on the combined
sample of male and female plots with plot-size dummies and gender-plot interactions. With
the inclusion of the nonhead dummy in this specification, the gender coefficient increases
in magnitude because it is picking up the differential in yields between female-managed plots
and those under the control of the household head. Thus, both the gender and the status of
the plot manager moderate the effect of the inverse size-yield relationship associated with
these plots.
Gender-Differentiated Input Intensities

Allocative efficiency, a basic condition of pareto efficiency, is achieved by equating
the marginal value product of inputs used in production to their unit costs. Thus, allocative
inefficiency stems from a failure to use profit maximizing levels of inputs. Cultivators
producing under constant returns to scale who are confronted with similar production
technologies and factor input prices should apply inputs in an equally intensive fashion.
Thus, assuming that these conditions hold, efficiency in production implies that input
intensities should be equalized in equilibrium across male- and female-managed plots within
the same household. Access to factor inputs should not be determined on the basis of one’s
gender.

Table 4a displays estimates of the labor intensities used per hectare on all of men’s
and women’s fields. These results corroborate Udry’ s findings: With the exception of female

labor, all labor inputs on a per hectare basis are used much less intensively on female-

26

managed plots in the same household. Women allocate more of their own labor--about
130% more hours--to their own plots than to those in the household managed by men.
Conversely, women farmed plots with less of every other type of labor input. On average,
women’s plots received 71.4% less hours of male labor per hectare than men’s plots in 1989
as displayed in column 3. The household’s children contributed 33.5 % less hours of their
labor to female plots. Animal traction labor, which is most often combined with male labor,
was provided 40% less per hectare on female plots.23

Moreover, the negative differential in the allocation of labor inputs revealed on
women’s plots could affect both the level and timeliness of application of other inputs. Thus,
these disparities could have a negative impact on both the marginal productivity of labor
used to complete certain types of agricultural tasks, and of other inputs used, such as
fertilizer and seed.

Since 1960, animal traction has been considered one of the most important catalysts
for productivity growth in the country and was the agricultural technology most singularly
adopted by cultivators throughout the Peanut Basin (Kelly, et a1 1996). Although originally
considered as a technology to raise growth via extensification, the real gains come from using
animal traction to increase intensification by applying other inputs more efficiently. Animal
traction can be used throughout the agricultural cycle - beginning with land preparation tasks
and finishing with harvesting - to increase the marginal productivity of other inputs. For
example, animal traction increases the effectiveness of fertilizer and manure applications.

Although used sparingly in Senegal, greater benefits from the application of fertilizer and

 

23

Animal traction labor is comprised of both human and animal labor inputs, and thus
is reported jointly.

27

 

 

28

82. 82 82. 82. ...-2. 82. 8.. 82. 2.l.22 ...-I... 2822
mm. 8. mm. 8. .2. 8. ..2. mm. .2
8...... 82 8...... 2...: 822.....8 8...... 8.2 8.82.5 .2...
8...... 2.8 822.22 8...... 222 822.22 8...... 82 8......82 8...... ... 2 822.222 2222.... m:
822.52-...
2:. 2:. 8.2 8.2 .22. .22. 222 ...22 m... .82.... :82
.22. 8.... .3. .. .82. .22.... .8... .22....
...... . :22 82: :28 2222 8...- 222 2.2250
.222. .22.. .. ...... .8...
22....- 22... .- 822.- 2222- 2223 an
.222. .85 .222. .222.
2.2..- 822- .2....- 8.8- 2223 ....
.22... .22. .22. .82.
222..- 8. 8- 22: .- 2......- 2..23 ......
.....2. .82. ....2. 82...
8. . 2- 3.2.- ..22- 8..- 2223 ....N
.onw “CE
.2... 82 .. .82. .22.... .222. .222. ..2: 82.2.
:2.- 82- 8.8- ..22- 2222. 822- 8.8 822 .2220
.N. ... .m. ... .N. ... .8. ...
22.8: .2. 228...: 3.. 9.88: 3.. 288: .2.
22.2.; .92.... .83 2...... .22.... 222 8...... 22......

 

.288. ...:u ...... 22.83:. 3.2... .2128.
82.25.... 2...... .52... .2 2.8.... .82.... 2.0
.... 2......-

manure could be obtained if these substances are worked deeper into the soil than possible
by using only human labor to move and spread the inputs.

Animal traction also enables cultivators to complete agricultural tasks within shorter
timeframes and thus follow prescribed dates that fall within narrow windows of opportunity
to achieve optimum yields. Combining animal traction with seeding tasks ensures that plots
will be seeded within the suggested timeframe for planting. Research conducted in Senegal
found that peanut yields are extremely sensitive to planting dates and decrease for each day
that seeding is delayed beyond the first seasonal rain. When cultivators use animal traction
to perform harvesting tasks they can reduce the risk of peanut crop failure. With the
introduction of a short-season peanut variety, it has become necessary to harvest the peanut
crop immediately after it has matured because it will regerrninate if it is exposed to additional
rain. Using animal traction enables cultivators to harvest their fields more quickly.

Table 4b reports input labor demand estimates for only monocropped peanut fields.
Labor use is significantly differentiated by gender of the plot manager. On average, female
cultivators receive 75.1% less male labor per hectare and 37.3 % less animal traction labor
per hectare than male cultivators. Even more telling, the household’s children spend
significantly less time on women’s plots than on men’s plots. Possibly devised as a strategy
to compensate for these other labor deficits, women allocate 106.6% more labor hours per
hectare to their own fields than to those managed by men in the same household.

Table 4c reports estimates of labor intensities for millet and sorghum fields in 1989.
Gender-based differentials of male and child labor hours on these fields are similar to those
found on peanut fields. However, animal traction is used even less intensively on women’s

millet and sorghum fields than women’s peanut fields. On average, 73% fewer labor hours

29

 

 

 

.22 .22 .22 .22 .22 .22 .22 .22 .22... .2 .3522
22 22 2... 2... 2... .... 22. .2 ....
8...... 22 8...... :2 8...... ...2 8...... 2...2 3.5.5.2. ......
8...... ....2 82.2.8.2 8...... 2.2 82...... .2 8...... .. .2 822.222 8...... 22.2 822.222 8.5.2.... ....
0.2.2.2 ..
8.2.. 2.2.. ...22 ...22 222. 2.22. .2.. .2.. .... .2.... .2 28.2
.222. .222. .22... .22... .8... .2.... 8...... 8....
2....2. ..222 22.2.2 2228 ......2 2......2. 2.2.. 2... .2288
82.... .22.... ..2... .222.
.. .22- 2.222.- 8.2...- 2.2..- 22...... ..2
.222. ....2. .2. .... .82.
..222- .22...- 22.28- 22....- 2.......... ....
.. ..2. 82.2. .222. .222.
.2. . 2- 2....2- ..2. .2- 8.2- 22...... ..2
.22... ..22. .28... .....2.
2......- ....22- 22.2- 22.2- 2.22.... ....2
”on? «CE
.8.... .2. .2. ....2. .8... 82... .222. .222. .22..
..2..- 222- ......2- ..2.22- 22.8- 22.2- .2... 22.22 .2250
.2. ... .2. ... .2. ... .2. ...
82.8... .2... 288.. .2. 2832...... 288: .2.
cores-.... BEE< .8...... ...—EU .33 0.2.). 3...... 22:5...

 

.28.... 22.88% .2.... 328.222...

momumE—Oufl— «BA—n: hag—a..— .uc mouaamumm muooum ficfimrfl WHO

5. 03.2.

30

of animal traction were allocated to women’s fields than men’s fields in the same household.

As suggested above, the benefits of animal traction may be indirectly transmitted
through the more efficient application of other inputs such as fertilizer. In particular, cereal
crops are more responsive to fertilizer than are peanuts (Kelly, 1993). Therefore, the yield
return that millet and sorghum cultivators experience from using fertilizer would be less for
those who are female because they use significantly less animal traction labor.

Table 4d reports estimates of non-labor input intensities for all fields in 1989. On
average, women are allocated plots that are 65.8% smaller than ones farmed by men in the
same household. While a negative plot size - yield relationship was found, women produce
less output for every size category than men when controlling for plot size effects.

The estimates show that, on average, women seed their plots less intensively than
males in the same household. Access to certified seed stocks, particularly for peanut
farming, has been identified by Senegalese farmers as a critical constraint in increasing
yields. Until 1985, seed was formerly distributed in a program that enabled farmers to
exchange part of their harvest in the current period for seeds to be used in the following
- agricultural season. After the program was discontinued by the government, cultivators
could only obtain new seed with either cash or credit purchases. This change in policy forced
some cultivators —namely women and unmarried sons living in the household within their
extended family networks—to switch from growing peanuts as a cash crop to cereals. Thus,
the estimates of seeding densities would under report the differential impact that women
would experience in obtaining seed. For those women who would have been most
constrained in their capacity to obtain reliable peanut seed, it is likely that they were forced

to farm alternative crops. The estimates do not capture this severe form of constraint.

31

 

 

222 222 222 222 222 222 222 222 8.214.... .2. 2IIIE=z
2... ..2. .... .2. 2... .... 8... .... .2
8...... ..8.8 8...... 22.2. 8...... 282. .8.... 2...2 8.8282. 8...
8...... 22.. .22....22. .8.... 22.. 882.2... 8...... 82.. 88......2. 8...... 2... 88......2. 8.8.8.2. ....
8.8.8.2-.
22.. 22.. 22. 2.2. 222. 222. ...22 ...22 .... 82... 8 282,.
.22. .. .22.... ..2. .. ..2.... ...... .22.... ...... ......
...2... 8.2 ....82. 8.2 22.222 2.22 22... 2... 82.8.6
.222. .222. .82... 82.2.
.... ..- 8222 . - 2.222- 282.- 2......=.. ..2
.222. ..22. .222. .. . .2.
.....22- 82......- 82.222. 222..- 2.......... ....
.222. .222. ..22. 8.2.
88222.. ..222 .. 2...8 . - 22.2.... 2.8....U ...2
8.2. .222. .28.... ..22.
22...- 22.. . .- 22.2...- . . .22- 2.8.8.. 8.2
.uNmm .07.
82.2. .22... .2.2. .22.... ..2.... .2.. .. .2.... .2.2.
8. 22- ..2.- 22.8..- 2...8- 2.2...- 2.. . .. ......2 ..222 82......
.2. ... .2. ... .2. ... .2. ...
258: .2. 98.8.. .2. 288.. .2. 0.8.8: .2.
8.82... .22...... 822-. 2...... 822-. D....... 8.3 0.2.5...

 

amuuubm 22.0253... 229.... SEE-8m .52 22...).Iawm—
8.2.2.8:— 22.522.21— uo 283:..2m 2.3.5— .5....— mac

9.. 033,—.

32

 

 

222. 22. 2.2. ...-2. 222. ...2. 222. 222. 222. 2...»... ... .22...-22'
22. 22. ..2. 22. 22. ..2. 22. 22. 22. ...
8...... .2...2 8...... .22 8...... 22.2 8......2222 8.88:2. ......
8...... .22 82.2.2.2 8...... 22.. 82......2. 8...... 22.2 8......222 8...... 22.2 8.....8. .2 8...... 22.2 2“.5522. 2:
8.2.6.2-...
.... 2.. .... .... 2... 2... 2.2.. 2.2.. ..2. 22.. .28.... :82
82.... .22. .. .2.... .... .... 82.... .22.... ...2... .... .... .... ....
2.... 22. 2.....- 2..... ......- 2... 2....2 3.2- 22.. 2.2.2.82.
.22.... .222. .222. .22....
22... 2..... 2..... ....22- 2.5.2.. ..2
.2 . .2. 82.2. 82.... .....2.
... ... 2..... ...... ....22- 2.2.8.. ....
.82. .22. .. .8... ..22.
2..... ...... 2..... ..2. .2- 2.2.2.. ..2
..2... .2.... .222. .222.
2..... ...... 2..... 2. ...2- 2.2.2.. ....2
62.2 ......
.222. .222. 82.... .22... 8...... .22.... .22... 82.2..
2.....- 2. .- ...... 2.....- ....o. 22...- 22....- .22- 22...- .2250
.2. ... .2. ... .2. ... .2. ... ...
.22... 2. .. .28.. 2. .. .23... .. .. .22...... .2. 22.. .2298...
022%.... 52:33... 95:22 39m 22¢. SE

 

Amuuobm 9:0 15. 22.02535 229..— =<Iaw3

3. 2...;

momummEOun-m wag—=— hGQflI—IEGZ HQ mauaamumm 390th Coflmrfl WHO

33

The table also reports that fewer women than men utilized manure but this
differential was insignificant. The estimates suggest an increased usage of fertilizer by
women over men, but the coefficient is not significantly different from zero. Fewer women
apply fungicide to their seeds than do men which could have a negative impact on the level
of seed germination and, thus, on yield outcomes. Although the gender differential is
significant, the point estimate is extremely small.

The use of manure and fertilizer is not widespread in Senegal. Therefore, it is not
possible on the basis of these estimates to contend that the marginal product of fertilizer is
decreasing and that further benefits could be exploited by redistributing these inputs to
women’s fields planted to the same crop in the household. These estimates do not support
Udry’s findings regarding fertilizer application on fields cultivated in Burkina Faso.24

Table 4e reports estimates of non-labor input use on peanut fields in 1989. The
estimates of plot size allocation and seed density provide telling evidence of resource
disparities by gender. On average, women farmed peanut plots that were 32.1% smaller than
those farmed by men in the same household.

Peanut seed, however, is considered by most Senegalese as the critical resource
constraint, and having access to quality seed is the key to increasing peanut yields. New seed
must usually be purchased either with cash or credit. Alternatively, peanut stocks fi'om one’ s
own harvest of the previous season are used as seed during the next one for those who lack
the means to obtain new seed. But the productivity of stored seed does not match the new

seed that can be purchased from the government. Seed density per hectare is also inversely

 

24

Udry estimated a fertilizer coefficient of -16.33 kilograms per hectare and a t-statistic
of -2.54 for female cultivators. The mean of the dependent variable was reported as 7.78
kg/hectare, conditioned on use greater than zero.

34

 

 

35

222 222 222 222 222 222 222 12...... 2 .....ma..l..z
22. 22. 22. 22. 2... .22. 2... ....
8......2222 .22.... ...... 8...... 22.2. 3.5.2.... ......
8...... 22... 822.22.. 8...... 22.. , 82.2.22. 8...... .2.. 82......2. 8...... 22.. 8.2.2:... .2:
momumtmum .2—
..2. ..2. .... .... 2.22 2.2.. 22. .2.. 2...... 2.. 52.2
.22. .. .222. .... .... ...... .. ..2. .22. .. 82.2.
...... . .....- .... ...... 2.....2 .22.. 2.2.2.8
.2...... .22.... 82.2.
22... ...... .....22- 2.222... ..2
.222. .2... .. ...22.
...... 2..... 3.2..- 2.222... ....
82.2. ...2... .222.
22... 2..... 22.2..- ..222... ...2
.222. .22.... ......2.
2.... ...... ..2.22- 2.2.2.... 22
.onm .07—
.22.... ...22. .22.... .22.... .22.... .22... ...22.
8.... 2 . .- ...... 2..... 22.22- 22.2- .2...- .22......
.2. ... .2. ... .2. ... ...
.2.... 2. .. .22... 2. .. .898: .2.. 22.. $22.82..
022?: 22:22 Bow 28< ......

 

.3822... 2.2.2223... .2...... .2......7222.
8.2.2.23:— 259: .8...—4-5.7. .... 28.25.23— 2.8.2.:— 3......— mac
.... 0......-

 

 

 

222 222 222 222 222 222 222 222 222 22.2... 22 2222.22
..2. 22. 22. 22. 2... 2... 2... 22. 3.. .2.
:22. 22.2 8...... 22.. 8...... 22.2 .2.... 2......an .2...
82.2.22... 82.22... 82.2. .2.2 82.2...22 82.2. 22.. 822.22.. 82.2. 2.... 822:2. 82.2. 22.. 22.2.2.22. 5.
motmtmumum
2.. 2.. .2. .2. 22. 22. 2.2 2.2 22.. 22: 2...... 22 222$.
.222. .2.2. .222. .2.2. .222. .2.2. .22... .222. .222.
2. .2- 2.2 22.2- 2.2 22.2- 2.2 22.2. ...2 22.. .22...-22..
.222. .22... .2.... ..2.2.
2.2 22.2 22.2 22.2.- 2.222... ..2
.22... .222. .222. .222.
22.2 .2.2 2.2 22.2.- 2.2...22 ....
.2.2. .222. .222. .222.
22.2- .2.2 22.2 22.2.- 2.22.22 ..2
.2.2. .2.2. .2.2. .222.
22.2- 22.2 22.2 .....2- 2.2.2.... ...2
nufim «CE
.22... .222. .222. .222. .222. .222. .222. ...22. .222.
22.2- 22.- 22.2 222.- 222.2- 2.. 22.2- 22. 22.2- .22......
.2. ... .2. ... .2. ... .2. ... ...
.2222 2. .. .2222 2. .. .2222 2. .. .22...... .2.. 22.. .2258...
2203...... .22.—Eu... 2:522 Bow 22¢. ......

 

goo-cm— anomacl. Eaawuem .25. SEE-tama—
2033525 :3:— ...Eadéez .2.. 2225.22m— vacuum— ..oxa ma—O

.3. «Ea-n.

36

related to the amount of labor time required to weed peanut plots and it is positively related
to moisture content in the soil. Therefore, the density with which plots are seeded directly
affects labor efficiency, soil quality, and peanut yield outcomes.

Compared to males in the same household farming peanuts, women seeded their
plots with 25.5% less kilograms per hectare. Although these results indicate that women farm
peanuts with differential levels of seed input than males living in the same household, they
do not capture information on whether men and women use seed that is different
qualitatively. It is plausible that women would be more credit constrained relative to men
and would therefore be more likely to plant seed acquired from previous harvests than from
new purchases.

Table 5a provides estimates of labor inputs utilized on all fields during the 1990
cropping season accounting for household and crop effects. Resource allocation follows the
trends patterned in 1989. On average, women allocated more of their own labor (126.8%
more labor per hectare) to their own fields while men allocated less (35.9% less labor per
hectare). Women also farmed less intensively with child labor (20.9% less labor per hectare)
and with animal traction ( 19.8% less labor per hectare) than men in the same household.

Table 5b provides estimates of labor inputs utilized on peanut fields in 1990.
Although the same general pattern of gender-differentiated resource allocation is evident,

women seem to be allocating even more hours to their own plots while other types of labor
is applied to women’s plots in a less discriminating manner. On average, women applied
130.4% more of their own labor hours per hectare to their own plots, but men applied only
1 1.5% less hours per hectare to women’s plots. Regarding allocations of child labor and

animal traction labor per hectare, the average differential is positive for women’s plots but

37

 

 

22.. 22: 22: 22: 22.. 22.. 22.. 22.. 22.22. 22 .22e2z
22. 22. 22. 22. 22. 22. 22. 22. .2
.222. 22.2 .222. 22.2 .222. 22.2 .222. 22.2 22.52.22. .2.2
.222. 22.2 .22.2.22.2 .222. 22.2 .22.2.2..2 .222. 2.2 .22.2.22.2 .222. 2.2 .22.2.2..2 22.22222. ::
motmtmumum
2.22 2.22 2.22 2.22 2.22. 2.22. 2.22 2.22 22. .222. 22 222.2
.222. .222. .2.2. 22.2 .222. .222. ..2.2. .222.
222. .2 22.22 22.22. 22.22. 22.22. 22.22 22.22 22. .2 2.2.2222.
.222. .22.... .222. .222.
222.22- 22.22- 22.22- 22.22- 2.22.22 22
.222. .222. .22.2. .2. .2.
2222.- 2.22- 2 . .22- 22. . 2- 2.22.22 22
.222. .222. .222. .222.
2222.- 22.2.- 22.22- 22.2- 2.22.22 22
.2 . .2. .222. .222. .222.
2222.- 22.22- 22.2.1 22.2.- 2.22.22 22
EN? ~22—
.222. .2.... .222. .22... .222. .2. .2. ..22. .22.. ..
222.2- .2.2- .22.- 22.2- 22.22- 22.22- 22.22 22.22 22222..
.2. ... .2. ... .2. ... .2. ...
222.22: 22.. 222.22: 22.. 222.22: 22.. 222.22: 22..
22822-2 ...—£52. 2092-. 2.2.0 232-. 2.22 232-. 2.2622.

 

.3222... 22G 222 2.22222... 22.2... .2222.

22.2.2222... .2...: 2222-. 22 22.2.2222. 2.2222... 22...-2 .225

am 2.2—uh.

38

 

”2.1

 

 

22.. 22.. 3.. 22.. 22: 22: 22.. 22.2.2 22 222.222
22. 22. 22. 22. 22. 22. 22. 22. .2
.222. 22.2 .222. 22.2 .222. 22.2 .222. 22.2 22.52.22. .2.2
.222. 22.2 .22.2.22.2 .222. 22.2 .22.2.22.2 .222. 22.2 .222. . 2.2 .222. 2.2 .22.2.22.2 2222.222. .2:
momumtmumun—
2.22 2.22 2. .2 2. .2 2.22 2.22 2.22 2.22 2:. .222. 22 222.2
.222. .222. .222. .22. .. .222. .22. .. .222. .222.
22 . .22 22.22 .2.22 22.2 .2.22 22.22 22.2.- 22.2.- 222.228
.222. .222. .222. .2. .2.
222.2- .2.2- .2.2 22.2.- 2.22.22 22
..22. .2.... .2. .2. .222.
222.2- 22.2.- 22.2 22.2.- 2.22.22 22
..22. .222. .222. .22...
222.2- 2 . .2- 2. .2- 22.2.- 2.22.22 22
.22. .. .22... .222. .2.2.
222.2- 22.2- 22.2- 22.2- 22.2.22 2.2
.8? “ca
.. 2.2. .222. .222. .222. .22... ..22. .222. .222..
222.2 22.2 2.2 22.2 22.2- 22.2.- 22.22 22.22 222220
.2. ... .2. ... .2. ... .2. ...
22282: 22.. 2.522: 22.. 8822: 22.. 222.22: 20..
22.82...- _2E.:< 2022-. EEO 22.2-— 2.22 232-. 2.2822.

 

dodge-cm .2.2—02:23. 2..—o..— «sauna—lama.

magma—0a:— HSA—E— hanaw— HO mOuQEmenm— such—m— QOXFD mu—o

am 22.2-F

39

the estimates are not significantly different from zero.

Estimates of labor inputs allocated to millet and sorghum fields in 1990 are presented
in Table 5c. Resource utilization followed the 1989 pattern. On average, women farmed
more intensively with their own labor (85.0% more labor hours per hectare) and less
intensively with male labor (47.8% less per hectare), child labor (36.8% less per hectare) and
animal traction (40.2% less per hectare).

Table 5d presents estimates of non-labor input use on all fields in 1990. On average,
female cultivators farmed significantly smaller plots than men in the same household. The
average area of women’s plots is about 60% less than the area of men’s plots. Women
farmed less intensively with the other inputs reported in the table. However, the estimated
gender-differential of resource use is either not statistically different from zero such as for
seed and fertilizer use, or overall use of the input for both males and females is low such as
for manure and fertilizer.

Table 5e presents estimates of non-labor input use on peanut fields. On average,
women utilized greater quantities of seed than men but the differential was not significant.

In contrast, more men applied quantities of manure, fertilizer, and fungicide but the
differences were insignificant. In 1990, the proportion of peanut cultivators who applied
fungicide (52%) was much larger than the proportion of users in 1989 (19%).

Table 5f presents estimates of non-labor input use on millet and sorghum fields in
1990. The findings are similar to those discussed above for peanut fields. On average,
female cultivator’s plots were 58.6% smaller than male cultivator’s. Women seeded their
Plots more intensively than males, but the point estimate is not significantly different from

261‘0. On average, fewer women than men applied manure or fungicide but overall use is not

40

 

 

222 222 222 222 222 222 222 222 22.2.2 22 .22el2z
22. 22. 22. 22. 22. 22. 22. 22. .2
.222. 22.2 .222. 22.. .222. 22.2 .222. 22.2 22.22222. .2.2
.222. 22.. .222. . 2.. .222. 22.2 .22.2.22.2 .222. .2.2 .22.2.22.2 .222. 22.2 .22.2.22.2 22.25.22. :2.
motmflnum .2—
222 2.22 2.22 2.22 2.22. 2.22. 2.22 2.22 22. .222. 22 222.2
.222. .222. .222. .222. .222. .22... .2. .2. .222.
222.22 2.22 22.222 22.2.2 22.22. .2.22. 22.22. 2.2.. 2.2.2222.
.2.2. .22... .222. .222.
222.22- 2.22- .2.22- 22.2- 2.22.22 22
.222. .22. .. .2.2. ..2.2.
222.22- 22.2.- 22.22- 22.2- 2.22.22 22
.222. .2.2. .22... .222.
222.22- 22. .2- 22.22- .2.2 2.22.22 22
.222. .22. .. .222. .222.
222.22- 22.22- 22.22- 22.2- 2.22.22 2.2
.onm .03
..2.2. .22... .22... ..2... ..2.2. .2. .2-. .222. .222.
2222.- 22.. .- 22.22- 22.22- 22.22- 22.22- 22.2. 22.2. 22222.2
.2. ... .2. ... .2. ... .2. ...
222.22: 222. 222.22: 22.. 222.22: 22.. 222.22: 22..
20.8222 .2E.=< 22.2.24 EEO ..2.2.. 2.2.). 2092-. 22:22...

 

.2.22222. 2.22222... 22.2... 2.222222 222 2.22222.
222.222.... .2...: 2222-. 22 22.2.2222. 2.22.222 22...-.— 242.

22 2.222

41

 

 

 

22.. 22.. 22.. 22.. 22.. 22.. 22.. 22.. 22.. 222.222.222.27.
22. 22. .2. .2. 22. 22. 22. 22. 22. 22.
.222. 22.2 .2.2.22. .222. 22.. .2.2. .2.2 2.2.2.522. .2...
.222.. ..2. .22.2.2. .22.2.22. .22.2.22. .22.2.22.2 .22.2.22. .222. 2.2 .22.2.22.2 .222. 22.2 22.2.2222. ....
22.22.28m -2.
22. 22. 22. 22. 22. 22. 2.22 2.22 22. .22... 22 222s.
.22. .2 .222. .222. .222. ..2.2. .222. .2.2. .222.
22.2- 2. .- 22.2- 2.2 22.2- 22.2 22.2 .2.2 292220
.222. .222. .222. .222.
2. .2 22.2 2.2 22.2.- 2.22.22 22
.222. .22... .222. .222.
2. .2 .2.2 22.2 22.2.- 2.22.22 22
..2.2. .22... .22.22 ..2.2.
2. .2 22.2 22.2 22.2.- 2.22.22 2.2
.22. .2 .22. .2 .22.22 .222.
22.2 .2.2 22.2 22.2- 2.22.22 .22
”ONE go:
.222. .222. .2. .2. .2... .2.2. .222. .222. .222. ..2.2..
222.2- 22..- .222- 222.- 222.2- 22.- .22.2- 22.2- 22.2- 222220
.2. ... .22 ... .2. ... .2. ... ...
.2222 2. .2 .2222 2. .2 .2222 2. .2 .2522: .22 22.. .22.2.22.2.
0222.32... 202.55... 0.5.83. Bow 22< 8.2.

 

Afloat-r.— mEU .252 22.02535 2.20.2— .2453
202.252:— 39: 22.22—-287. no 20:22:22”.— fluubm can; mac

em 2...:-

42

 

 

222 222 222 222 222 222 222 222 222 22.2.2. 22 222.222
22. 22. 22. 22. 22. .2. 22. 22. 22. 22.
.22.2.22.2. ..2.2. 22.. .222. .2.2 .222. .2.. 22.82.22. .2.2
.22.2..2.2. .22.2.22.2 .222. 22.. .22.2.22. .222. 22.. .22.2.22. .222. 22.. .22.2.22.2 .222. 22.. 22.2.22... I:
. 22.22.222m -2.
22. 22. .2. .2. 22. 22. .2.22 .2.22 22. 22: .222. 22 222.2
..2.2. .222. .. ..2. .. ..2. .222. .2.2. .222. .222. .22. .2
2. .2- 222.2 .2.2 222. 22.2- .2.2 22.22. 22.2. . 222.2 222.2220
.222. .22... .22.22 .2....
22.2 22.2 22.2 22.2. 2.22.22 22
.222. ..2.2. .22... .22...
22.2 .2.2- 22.2 22.2. 2.22.22 22
..2.2. .222. .22.: .222.
22.2 .2.2- 22.2 22.2- 2.22.22 2.2
.222. .. ..2. .222. .222.
2.2 22.2- 22.2 22.2 2.22.22 222
”onm “GE
.22. .2 . .2.2. .222. .22.: .222. .222. .222. ..2.2. .222.
22.2- 2. .2- .2.2- .2.2- 22.2- 22.2- 2.2 22.2- 22.2- 222220
.2. ... .2. ... .2. ... .22 ... ...
.2222 2. .2 .2222 2. .2 .2222 2. .2 .2522: 222 22.2 .22.2.22.2
0.2.0.32... 52:33.... 05:22 Bum 22¢. 2o...

 

.3222... 2.2222225 22.2.2 222222.222.

magmas—0:: «59—: haa—RH—nn—OZ MG muuflamemm Quack—m— mvvkmrfl mac

em 035—-

43

 

 

222 222 222 222 22.. 222 222 222 222 22.2.2. 22 2222.42
22. 22. 22. 22. 22. 22. 22. 22. 2.2. 22.
.222.. ..2 22.2.22. 5.2.22. .22.2.22.2 22.22222. .2.2
8.2.23. 8.2.22... .22.2.22.2 .22.2.de 82.2.82 .22.2.22.2 .22.2.22.2 .22.2.22.2. .22.2.22.2 22.5.25 ..2.
22.22.2222 -2.
2.. 2.. .2. .2. 22. 22. 22.2. 22.2. 22. 22: 2222. 22 2222
.2222 .2.22 ..2.22 .2.22 .2222 .2.22 .2222 .2.22 .22.:
22.2- 2.2 ~22- 2.2 2.2- 2.2 22.2.- 2.2. 22.2 22222220
.2222 .22. .2 .22.: .2222
...2 22.2 2. .2 22.2. 2.22.2.2 22
.222 .2222 .222 .2222
22.2 .2.2 . ..2 22.... 2.22.22 22
.22.: .22. .2 ..22 .222
22.2 22.2 22.2 22.22 2.22.2.2 22.
.22.: ..2.22 .2222 .2222
22.2 22. 22.2 22.: 2.22.22 am
.2. .22 .2222 .2222 .2.22 .2.2...2 .2222 .2222 .2222 .222
22.2- 2.2.2- .2.2 2.2 N. .2- 2.2- .2.2 22.2- 22.2- A 222220
.N2 ..2 .22 ..2 .22 ..2 .N2 ..2 ..2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .2522: .22 2.2.2 .22.2.22.2
222%....22 23:22.2... 2.2.2.2 222w 222.2. 22.22

 

#22222... 2.2222225 22.2.2. 22...: 222 EEwSmug...
magma—32: 2..—flu acnadnfloz .3 8353mm 399cm Quinn mu—C

.«m 03.2%

44

high, nor is the fungicide estimate significantly different from zero. The estimates show a
slight increase in the number of women using fertilizer but this is not significant.
Resource Allocation and Plot Size Effects

If it is the case that women are systematically allocated plots that are inferior relative
to those cultivated by men, then differences in the allocation of resources based on gender
of the plot manager could be justified. Comparing the ‘gender-effect’ across different yield
specifications may provide some evidence that systematic differences in plot quality by
gender does occur. Unfortunately, the sc0pe of this survey did not cover issues related to
soil productivity. Therefore, the only information available in this data related to plot
characteristics is size.

The first columns appearing in each of the crop specifications in Tables 2a and 2b
exclude the plot size quintile dummys from the specification. In general, when these
dummys are not included the estimate of the absolute value of the gender coefficient
becomes smaller. This would suggest that characteristics particular to women’s plots have
a positive effect on their yields. If women are allocated the same plots over time, then these
quintile dummys may also be capturing information about soil characteristics.

Similarly, column (1) in each of the input demand specifications controls for
household and crop effects but does not control for plot-size effects. As discussed above,
if unobserved plot characteristics are correlated with other factor input intensities then the
plot quintile dummy is endogenous in the specification. Therefore, the gender coefficients
generated from the column (1) specifications should be interpreted as the total gender effect.
With the exception of female labor use estimation, the gender differential is reduced when

plot dummies are excluded from the specification. The gender coefficient in the column (2)

45

 

specification is inclusive of plot-size effects. Smaller plots are farmed more intensively and
women are allocated systematically smaller plots. Thus, the two specifications provide an
upper and lower bound from which to estimate the effects of gender-based resource
allocation within households.

The intensive farming of small plots may suggest that relative size distinguishes
between communal or personal ownershipwithin a household. Large fields, for example,
may be reserved by the household head for the purpose of growing household food crops.
Although family members would be expected to contribute some of their labor to these
fields, it may be difficult to monitor whether these contributions are actually fulfilled or
whether some shirking of labor responsibilities occur. Therefore, it may be more likely that
small plots—designated as personal fields-would be farmed more intensively than large plots,
particularly with regard to the intensity of labor inputs. However, this trend would not
necessarily be sustained in the provision of non-labor inputs, because decisions concerning
these factors do not depend upon the coercion of others or other monitoring costs.

Gender and Household Status Differentiation in Input Allocation

One might potentially argue that the real differences in intensity of input use do not
stem from gender but from household status. In particular, resource decisions would be based
on whether a plot was managed by the household head and whether it was being used to
grow some portion of the household’s food. The household head controls fields designated
either as communal, which are used to grow cereals for consumption by the extended family,
or personal. The personal fields would be planted with crops intended to be sold for cash.
Thus, it is the communal fields managed by the head that would be receiving a major

portion of the agricultural inputs controlled by the family and all other plots-those managed

46

 

by women or by other males, namely unmarried sons residing in the family compound--
would be allocated significantly fewer resources.

Along this line of argument, discrimination in access to resource allocation occurs
between household heads and non-heads. However, one should remember that if household
members in Senegal operated in a cooperative manner, it wouldn’t matter whether a field was
being used to grow food or cash crops because either could be converted into cash or food
through the market, and the proceeds from all fields would be pooled. Thus, the pareto
efficiency condition should still hold.

Tables Al - C6, located in the appendix, report estimates of yields and labor and non-
labor input use while controlling for resources allocated to non-head males living in the
same household. All of the specifications reported in these tables incorporate a nonhead
dummy allowing for comparison of resource use allocation between female- cultivated plots
' and between those cultivated by male heads, and for comparison between plots managed by
male non-heads and those managed by male heads in the same household. Thus, we interpret
the gender coefficient in this specification as the differential in resource use per hectare
between female cultivators and male head cultivators, and the nonhead coefficient as the
differential in resource use per hectare between male non-head cultivators and male
cultivators. In contrast, the exclusion of the non-head dummy in the tables presented above
allows us simply to focus on the differential in resource allocation outcomes occurring
between males and females.

Tables A1 and A2 report estimates of yield per hectare for the entire sample and
selected commodities for 1989 and 1990, respectively. The gender differential throughout

these specifications is larger in magnitude compared to similar specifications without the

47

 

nonhead dummies (see tables 2a and 2b). These coefficients should be interpreted as the
differential in yields per hectare between plots managed by females and those managed by
male household heads. The nonhead dummies pick up the differential in yields per hectare
between plots managed by male nonheads of the household and by male heads. Thus,
productivity within the household is influenced by demographic variables.

Table Bl reports estimates based on 1989 labor input intensities for all of the
combined fields. Women provide slightly more of their labor to male head plots than to
those managed by male non-head cultivators. Comparing all of the other types of labor
allocations (male, child, and animal traction) between the Column (2) specifications
presented above and Column (2) specifications in the appendix, the disparity in resource
access widens and becomes even more pernicious for female plot managers. The non-head
coefficient indicates that differentiation in resource use exists between the plots controlled
by male heads and male non-heads. On average, male non-head cultivators allocate 36.1%
less male labor per hectare to their own plots and receive less female labor (-9.42%) per
hectare, less child labor (-3 l .7%) per hectare, and less animal traction
(-39.2%) per hectare.

Resource allocation of labor on millet and sorghum fields in 1989 exhibits the same
patterns found above. Table B2 reports these estimates. On average, women allocated more
of their own labor to plots managed by male heads than to the entire group of males, but they
still reserved more of their labor for their own plots. Male non-heads received 64.0% less
female labor per hectare on their own plots. More tellingly, male non-heads allocated 82.8%
less male labor per hectare to their own plots than to those managed by other males in the

household. On average, non-head male cultivators also received significantly less child labor

48

 

(-108.2%) and significantly less animal traction ( -90.7%) per hectare. Comparing the
Column 2 specifications for each type of labor reinforces the notion that women are allocated
less labor input hours per hectare compared to all men in the household, and the
differentiation only increases when restricting the comparison between female plots and male
head plots.

Table B3 reports estimates of labor input use for peanut fields in 1989. The pattern
of resource use disparity was similar for women while it was diminished for non-head male
cultivators. These results may provide the most striking evidence that male and female
conjugal units do not allocate resources efficiently. Restricting the sample to only
monocropped peanut fields allows us to compare resource allocation across cash-crop fields
which implicitly means personal fields. Women reserve more of their own labor for their
own fields, but the differential decreases when the comparison is made between plots
cultivated by male heads and those cultivated by females in the same household. The other
types of labor were applied significantly less intensively to female plots when compared to
all males or restricted to male-head plots, but the differential is reduced when compared to
the plots of all males in the same household.

The dichotomous pattern of male-female and head-nonhead household labor
allocation is maintained in 1990. Table Cl reporting results from all of the fields included
in the 1990 sample shows that females allocate more of their own labor to their own plots
than to those cultivated by household males, but the gap narrows when the comparison is
restricted to plots of the household head. In contrast, females cultivate less intensively with
every other type of labor than males in the same household. The estimates become even more

significant when the nonhead dummy is incorporated into the regression, thereby restricting

49

the comparison of resource use to between females and male heads of the household. On
average, male non-head cultivators also used less male labor, child labor, and animal traction
per hectare than the male heads, but the differential is smaller than what is found for women.
Similar trends of labor use are depicted in Table C2 in which the estimates are restricted to
fields planted with millet or sorghum.

Table C3 reports estimates of labor input intensities for peanut fields in 1990. These
results diverge from the typified pattern. On average, women farm with less male labor per
hectare but there is little differentiation in use of child or animal traction labor. The non-
head dummy is insignificant in all of these estimations which means that no differentiation
in labor use occurred between male non-heads and heads in the same household.

Estimates of non-labor inputs utilized on all of the fields in 1989 are reported in
Table B4. Both female and non-head cultivators in the same household farm significantly
smaller plots than compared to those of the household head. Women use substantially less
seed per hectare than all other males in the household with little distinction between non-
head males and heads. The differentiation in seeding intensity occurring between head and
nonhead plots is insignificant. On average, the frequency of manure and fertilizer use on all
household plots is low and any differentiation in use by household member is insignificant.
Although 20% of the plots report some use of fungicide, male non-heads and females in the
same household use substantially less than male heads.

Table B5 reports estimates of non-labor inputs for millet and sorghum fields in 1989.
On average, both women and non-head males farm significantly smaller plots than the
household head. Plot-size differentiation across these groups may not be surprising as most

of the millet and sorghum fields managed by the household head would be intended to

50

 

produce most of the cereals consumed by the family throughout the year. With the
exception of fertilizer, all other inputs are used less intensively by females and non—head
males than heads in the same household.

Non-labor input use on peanut fields in 1989 is reported in Table B6. Plot-size and
seeding density per hectare are significantly different for female cultivators than for others
in the same household. With the exception of land allocation, little differentiation in the
intensity of farming occurs between non-head males and household heads. The application
of fertilizer on peanut fields in 1989 was almost nonexistent. Thus, estimates could not be
generated for this input.

As we might expect, fields generally considered to produce the household’s primary
staple (millet and sorghum fields managed by the household head) receive relatively more
agricultural inputs than those designated as personal. This assumption is supported by the
finding that both male non-head and female cultivators farm millet and sorghum plots less
intensively than male heads in the same household. However, discrimination in access to
resources occurs for plot managers of peanut fields—but only for female plot managers. This
finding is not only counter-intuitive, but it reinforces the general finding of the data that
households do not allocate agricultural resources efficiently and therefore can not be
generalized as cooperative units.

The strong relationships found between plot size allocation and household
characteristics in 1989 remain consistent throughout the 1990 specifications. However,
many of the other relationships between non-labor resource allocation and plot manager
characteristic are generally weak, primarily because utilization of these is inputs is low. Table

C4 reports non-labor input intensities estimated for all fields in 1990 with household and

51

 

Im—

crop effects. Females and male non-heads employ less manure ( although overall use is low)
and less fungicide than male heads. The differentiation associated with use of seed or with
fertilizer is not significant.

Non-labor input intensities for fields mono-cropped with either millet or sorghum
are provided in Table C5. Land area is the only input for which critical differences in
allocation by gender or status of plot manager occur. Females and non-head males employ
less manure and non-head males employ less fungicide per hectare than male heads in the
same household, but overall use is marginal.

Table C6 reports estimates of non-labor input use on peanut fields in 1990. With
respect to land area, both female and non-head males cultivate significantly smaller parcels
of land than male heads in the same household. Interestingly, females and non-head males
seeded their plots more intensively than household heads but neither of the point estimates
were significant.

How Serious is the Misallocation of Resources?

It has been demonstrated that the manner in which households allocate productive
resources across individually managed plots is inefficient, thereby resulting in a loss of
potential output. But how serious is this loss and what can be done to remedy the problem?
Without the existence of well-functioning labor and land markets, one might expect that
yields generated by a household will vary according to its capacity to manage resources,
respond to risk and operate in a cash environment where credit is almost nonexistent. Such
an environment could lead to far greater variability and loss in production than what may
be associated with intra-household inefficiencies. Therefore, it’s important to evaluate the

relative efiiciency losses occurring within households and within villages.

52

 

A simple test can facilitate the comparison of yield variations resulting from
differences across households to those from within the household. Household dummies
capture the average dispersion in yields occurring across households. Such variation can be
attributed to the differentiated set of land, labor, capital, and finance entitlements acquired
by a household and its capacity for modifying this set through participation in existing
markets. Climate will condition both the household’s endowments and the responsiveness
of markets.

In the following table I present descriptive statistics of the household coefficient
matrix. By computing the interquartile range of these coefficients, I can compare the relative

variation in yields accruing across households to the gender-based variation within

 

 

households.
Table 6
Household Dispersion in Yields
1989 1990
25th Percentile 4843.06 -20516.80
50‘h Percentile 6144.51 -11014.86
75th Percentile 21165.87 6942.39
Inter-quartile Range 26008.93 27459.19

 

 

 

In 1989, the variation in yields across households is 26,0089 cfa per hectare, almost
three times the yield variation occurring between men’s and women’s plots. In 1990, the
dispersion is further accentuated. Household variation in yields is about four times the yield
variation occurring between men and women within the same household. Not surprisingly,

the gender-based variation is small—but not insignificant-- in comparison to the inter-

53

 

household variation. Fluctuations in weather combined with substantial variations in
Senegal’s agroclimatic landscape drive much of the variability in yields across villages and
across seasons.

Following along the dimensions that Udry used to measure the relative importance
of household and village production losses, I re-estimate the output supply equation to
determine the extent to which actual plot yields deviate from predicted yields if factors are
allocated efficiently across individuals within a household, and if factors are allocated
efficiently across households within each village. Gender is dropped from each of the
following specifications because the obj ective is to explore for all possible sources of factor
misallocation. Accordingly, although yield outcomes are sensitive to plot size, I do not
control for any plot-size effects because land area is strongly correlated with the gender of
the plot manager.

The specification is intended to measure the relative severity of factor misuse. First,
assuming that individuals make rational and efficient decisions about resource allocation
across the various plots under their control, I compare the deviation from predicted yields
occurring under individual control to those at the household and village.25 Theoretically, any
deviation experienced by individuals should be attributed to variances in plot-level
characteristics and would sum to zero if there was no unobserved plot characteristics. Thus,
the error term of this equation, em 0), should contain yield variation resulting only from plot-

specific risk or idiosyncratic plot effects—and not from factors related to factor misallocation.

 

25

Plots are identified with a code based on the manager’s relationship to the household
head such as first wife, mother, or head of the compound. In most cases, these codes will
serve to identify uniquely across plots managed within a household. However, if more than
one unmarred son resides in the household and more than one of them manages a plot, then
these managers will not be identified uniquely.

54

 

 

'1‘. a. nun.- II- '-

Second, the same equation is estimated with crop and household effects. The error
term in this specification, E he (h), comprises variation resulting from plot-level sources and
from inefficiencies in factor allocations across household members.

The third specification is estimated with village and crop effects. The sources of
variation captured in this error term, €12 (v), include inefficiencies in factor allocation
occurring across plots controlled by different households within a village, as well as the
sources identified above in the first and second specification.

Figures 1 and 2 report the results of a separate kernel density function estimate for
each of these error terms by year. The functions are graphed on the same scale so that the
dispersion can be compared visually.26

Inspecting the distributions of the error terms corresponding to the individual-crop
effect and to the household-crop effect in 1989, it is apparent that the latter is more diffuse.
This implies that there is more dispersion in yields occurring across plots managed by
different people within the same household than there is in yields occurring across different
plots but under the management of the same individual within the household. This notion
is reinforced statistically. I test the hypothesis that the individual-crop effect, (v J.), included
in the first specification is equal to the household-crop effect, (vh), in the second
specification. The test statistic for this comparison of regressors is distributed as F (400,

999)= 1.5 and has a p-value=0.0.

 

26

The kernel density uses the locus of points located around each point in the
distribution to estimate a separate density for every x along the x-axis. The points used in
each estimation are known as a band width and their weight in the function is based on their
closeness to the particular x being estimated. These band widths are similar to the bins
depicted in the histogram but minimize the potential bias that results from a static
representation of data. For further discussion, see Deaton (1997) pp 170-179.

55

K.‘ LIE..-

 

It is more difficult to compare the error distributions of the household-crop effect and
the village—crop effect because of the skewness in the latter, but the test statistic allows us
to reject the hypothesis that the household (vh) and village (v,) effects are similar for all
households residing within a particular village. The test statistic is distributed as F (180,
1399)=1.7 and has a p-value=0.0. Thus, less dispersion occurs in yields within the same
household than occurs in yields across different households within the same village.

These tests reinforce the notion that household allocation of productive resources
across plots are inefficient and the magnitude of the distortion is significant even when
compared to the potential misalignment of productive factors at the village level. Although
the underlying inefficiency caused by the dimension of gender in factor allocation is costly
in terms of the potential output that is not realized, other forces may be at work. The
statistical comparison within and across households considers all possible sources of factor
misallocation and thus is more general than the one used to determine whether inefficiencies
exist on the basis of gender.

Figure 2 compares results of yield dispersions from predicted yields across different
plots managed by individuals within a particular household, across different individuals
within a particular household, and across different households in 1990. From the figure it
appears that the error term of the household level distribution is more diffuse than the one
corresponding to the distribution of the individual-crop effect residual. However, based on
the test we cannot reject the hypothesis that these two distributions are similar. Therefore,
there is little difference between the dispersion of yields across plots controlled by an

individual and across plots controlled by different individuals within household.

56

 

.00003 —

.00002 “

1 .Oe-05 ‘

 

0—

Production Residuals, 1989

Person Level

Household Level

Village Level

 

 

-1odooo

Figure l

2oo'ooo 300060

0.1
A
o
0-
o
o
o

57

 

Production Residuals, 1990

.00004- /\

Person Level

   
  

 

Household Level

 

 

 

.ooooz -
Village Level
0 - 2 __
I l
-1ooboo 0 100000
Figure 2

58

In contrast to 1989, the greatest variability in yields occurs across different
households. The results depicted in figure 2 suggest that yields are more diffuse within
households, but statistically the hypothesis that similarity exists between the household-crop
effect and village-crop effect for every house in a particular village is rejected. The test
statistic is distributed as F (138, 1044)=2.4 and has a p-value=(0.0). Therefore, more
dispersion occurs in yields across households than across plots within the same household.
Factor misallocation across households may be attributed partially to the nonexistence of
village factor markets or to an environment in which these markets do not function well.
While more problematic for obtaining optimal yield outcomes than household inefficiencies,
it does not diminish the significance of the results presented above that productive resources
within the household are allocated based on the gender of the plot manager. As was noted,
this test examines all possible sources of inefficiencies and should be considered less
powerful when compared to the more specific finding of gender-based inefficiencies.

IMPLICATIONS

Kelley (et al, 1996) concluded that household yields in Senegal are differentiated by
the household’s access to and use of better soils in which to plant crops, household labor
(both human and animal), and cash to purchase more and better inputs. The results
presented here have demonstrated empirically that the differentiation in access to
agricultural inputs is further transmitted to individuals within households. Resource
decisions within the household are being made on the basis of gender. Although one might
argue that in an environment where resource constraints are particularly critical and some
field managers within a household are allocated inferior land and less inputs per hectare

because there isn’t enough to accommodate all household needs, should these decisions be

59

made systematically on the basis of gender?

Evidence suggests that a systematic allocation of plots on the basis of gender would
be inefficient. Both male and female cultivators apply exceedingly low quantities of
fertilizer and manure on their fields. Thus, to compensate for the deleterious effects of soil
nutrient depletion these households should be practicing crop rotation and fallow agricultural
techniques that replenish the nutrient content of the soil. Crop rotation can improve soil
structure, enhance permeability, and increase biological activity, water and nutrient storage
capacity, and the amount of organic matter (Gebremedhim and Schwab, 1998). In Senegal,
the planting of peanuts, which are nitrogen-producing, should occur on land formerly under
cereal production. Millet and sorghum are nitrogen-absorbing crops.

Female cultivators farming peanuts should be allocated different parcels of land
across time to spread the benefits associated with the production of nitrogen in household
land. If, for example, we find that women, on average, are allocated the relatively inferior
household plots within a given year, then we should expect to find women being allocated
relatively better household land during the following year. In a system of cereal-peanut crop
rotation practiced by households, it is conceivable that land reserved by the household head
for communal cereal production in a given year could be reallocated in subsequent years to
peanut cultivation. Therefore, in an efficient setting, we would expect the assignment of land
to be based on the particular crop grown. In a household where the availability of good soil
is limited, the household head should be rotating crop production to achieve maximum yield.
The gender of the plot manager should not factor into the land allocation decision.

Other inputs could be heterogeneous. Another omission that is suggestive of this data

is managerial ability. It is possible that women are inferior plot managers compared to men.

60

"Km

 

If this scenario were true, gender-based differences in yields could result from technical
inefficiency. I suggested above that the limited use of animal traction on women’s fields
could delay the timing of input applications or other events that negatively affect yield
outcomes on women’s plots. If household heads were aware of the differences in ability
existing between female and male cultivators within the household, they could allocate
resources on the basis of this difference. In contrast to a gender-based allocation scheme,
resource differences that occur because of heterogeneous ability could be considered
efficient. The household head could be minimizing potential productivity losses by
reallocating resources away from inferior plot managers.

If this is so, why would the practice of multiple plot managers be sustained? If
women are systematically weak managers, wouldn’t it make sense to use their talents
elsewhere? Women could serve as a pool of labor for household fields and be remunerated
for their services. Men, the superior plot managers, would make better use of women’s
labor. In exchange women could be compensated for their time. Thus, if significant and
systematic differences existed between the genders in plot management activities, the
utilization of women as managers would be allocatively inefficient.

The relatively high incidence of polygyny occurring within Senegalese households
suggests that the marginal contribution of women’ s labor is high. J acoby (1995) disentangles
the wealth effects from the shadow price effects associated with the demand for wives in the
Ivory Coast. The shadow price of a wife is defined as the marginal value of investment in
physical capital foregone to pay for the bride price of an additional wife combined with the
cost of retaining that wife, but offset by the marginal productivity of her labor on farm. He

finds that, conditional on wealth, men marry more wives when female labor contributes to

61

 

a larger share of agricultural income. In Senegal, women contribute to the household food
supply with the production and market sale of an important cash crop, peanuts.

Women’s labor is demanded but labor markets are non-existent. Thus, the family
intemalizes the absence of this market. The prevalence of polygyny suggests that men vie
for women’s labor through the marriage market. This bids up the cost, or bride-price, of
women and increases inequality among men. The ratio of available women to available men
in the marriage market is often cited as a factor that contributes to a woman’s relative
bargaining power within the household.27 23 Thus, some of the negotiation for household
resources between men and women may occur as part of a marriage contract. Wealthier
household heads bid away women with the promise of access to land, a productive asset that
generates a personal stream of income.

Farming a separate parcel of land and selling that output in the market offers women
more autonomy then if women were paid for their labor. Money is fungible. Less inclined
to contribute to other household expenditures, men may deduct the earnings paid to women
for their labor from other promised household transfer payments. In a farming situation, men
don’t really know what a wife makes from her sales and effort. They would be less able to
make these types of deductions.

Men may also benefit from an arrangement in which the responsibility of plot

 

27

McElroy (in Haddad, et al, 1997) refers to these factors as extrahousehold
environmental parameters (EEP’s), “threat-point shifters” that indirectly affect an
individual’s share of income within the household through the individual’s relative position
of power.

28

Goldman and Pebley (in Lesthaeghe, 1989) discuss three factors that determine the
available sex ratio in marriage: the incidence of polygny, the average gap in the maniage age
between men and women, and the rate at which widows reman'y.

62

 

management is shared with women and other household males. The household head may use
multiple plot managers as a risk management strategy. In exchange for labor provided on
communal fields, the household head allocates land to household members for their own
production. But he passes the associated costs of agricultural risk on to them as well. In the
event of a bad harvest season, plot managers absorb the losses from low yields. However,
if these individuals were compensated for their labor on the household head’s fields with
wages, they would still expect to be paid. Plot managers bear the costs of risk in a way that
paid laborers do not.
CONCLUSION

In a setting in which household members bargain for personal control over resources,
the systematic allocation of productive inputs on the basis of gender implicitly determines
income distribution within the family and may drive household yields away from its
production possibility frontier. Hence, production decisions within the household calculated
to influence the relative share of consumption accruing to household members are not
separable. Evidence from this research suggests that households are willing to sacrifice
potential efficiency so as to be able to alter the internal resource allocation process.

This research has provided evidence that Senegalese households have been unable
to achieve allocative efficiency in farm production and provided fiirther empirical support
to the findings of Udry (1996) and Jones (1983). The failure to meet this criteria suggests that
neither the unitary model nor the more general collective bargaining model sufficiently

describe the household resource allocation process in West Afiica.

63

 

APPENDIX A

64

 

 

 

 

 

65

222 222 222 222 222. 222. 22.2.2 22 222el2z
22. 22. 22. 22. 22. 22. 22
.222. 22.2. .222. 22.2 .222. 22.2 22.2.55 .2.2.
.222. 2.2 .22.2.22.2 222.2. 22.. .22.2.22. 222.2. 22.. .222. 22.. 22.52.22. :2.
22.22.2222 -...
w< 2o :82
.22.22 .22.22 .22.22 ..2.22 .22.22 .2...2
.2.2222 22.2.22. 22.2222. 22.22222 22.22222 2.22222 222222222
.2222 .2222 .2.22
2.22222- 222.222- .2. .2222- 2.22.22 22
.22.22 ..2.22 .22.22
2.22222- 22. .2222- 22.22222- 2.22.22 ...2
.22.22 .22.22 .22.22
.2.22222- 22222.2- 222222.- 2.22.22 2.2
.22.22 .22.22 ..2.22
22.2222 .- 22.222 . 2- 22.2222 . - 2.22.22 ....2
6&2 SE
..2.22 .22.22 .2222 .2. .22 .2.2 ..2.22 .2222222 2.22 2. .2
.2.2222. 2. 2222.- 222222.- .2.2222.- 22.2222.. 22.2222- 222.222
.2. .22 .22.22 .2222 .22.22 .2222 .2222
22.22222- 22222.- .2.2.222- .2222... 22.22.22- 2..2222.- 222220
222 ..2 .22 ..2 .22 2.2
5222.22.22.22 222222. 2226 ..<

 

2222.:m ..2-.0 ES 22.—22:2: 52.: 2.8222: ..2.— 2==2>2m ..2.22-.2.2..— ama—
_< 232,—.

 

 

 

 

hr...“
. 222 222 222 222 22.. 22.. 22.2.2. 22 222.222
22. 22. 22. 22. 22. 22. 22.
.222. 22.2 .222. 22.2 .222. 22.2. 22.22222. .2.2.
.222. 22.2 82.2.2.2 .222. 2.2. .22.2.22.2. .222. 22.2 .222. .2.2 22.22222. ..2.
22.22.2222 -2.
w< .22 52.22
.22.22 .22.22 .2 . . .2 .22.: .22.22 .22.22
22.222 . 2 22.22222 .2.2.22. 22.22 . 2. 22.22222 .2.22222 222222222
.22.22 .22.22 .2.22
22.222- 22.2222 2.22.2- 2.22.2.2 ...2
.. 22.22 .22.22 .22.22 6
22.22 .2 22.222. 22.2222- 2.22.22 ...2 6
.22. .2 .22.22 .2...2
.2.2222 22.222 .2. 2222- 2.22.22 ..2
.22. .2 .22.22 .22.:
22.2222 22.2222 22.2222- 2.22.22 2.2
222.2 ..2.2.
.22.22 .22.22 .22.: .22.: 22.2 .22.22 .2222222 2.22 2:2
22.2222- 22.2222- 2.2. .222- .22.2.22- 22.2222- 22.2222- . 222.222
.. ..22 .22.22 .22.22 .22.22 .22.22 . . 2.22
22222.- 222.22.- 22.2222- 22.2.22- 2..2.22.- 22.2222- 222.2222
222 . .2 .22 . .2 .22 2 .2
63.238352 23:22.2 22.80 =<

 

muoobm .22.—U .2...2 22.—22:2: 5.3 9:88: ..2.— 2.:923— .9632..— .223
S. 2..—ah

APPENDIX B

67

 

 

...-Im‘Hl...‘ II.

 

 

2222 2222 2222 222. 222. 2227 222. 222. 222.222.22.232:
22. 22. 22. 22. 22. 22. 22. 22. 22.
82.22.22 .22.2.22.2. 2222.222 82.2.2.2. 22.22222. 22.2.
.22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 82.2.2. .2 .22.2.22.2 22.22222. ..2.
22.22.2222 -...
2.22 2.22 2.22 2.22 ..22. ..22. 2.22 2.22 2:. .222. 22 2222,.
.22.22 .22. .2 .22.22 .22.22 .22.22 .22.22 .22.22 .22.22
22.22. 22.22 2.222 22.22 22.2. 2 22.22. 22.22 22.2 222222222
.22.22 .22.22 .222: .22.22
22.22- ...22.- 22.22.- 22.22- 2.22.2.2 22
.22.22 .22.22 .22.22 .22.22
22.22- 22.22.- 22.22.- 22.22- 2.22.2.2 22
.2.22 .22.22 .22.22 .22.22
22.2.1 22.22- 22. . 2.- 22.22- 2.22.2.2 ..2
.22.22 .22.22 .22.22 .22.22
2. .22- .222- 22.22- 22.22- 2.22.2.2 2.2
ENE 20:
.2 . .22 .2.22 .22.22 .22.22 .22. .2 .22.22 .2 . .22 .222....22 2.222. .2
22.2.- 22.2- 22.22- 22.22- 22. . 2- 22.2- 22.. 2222222
.22.22 .22.22 .22.22 .2222 .222: .22.22 .22.22 .22.22
22.22- 22.2.- 22.2- 22.2- 22.22.- 22.22- 22. .2 22.22 2222222
.22 ..2 .22 ..2 .22 ..2 .22 ..2
222.22: 2022 2.8222: 22.2 9.822: 22.2 22522: ..2.
2.2822..- _2E_=< 2.52-— EEU .5222»— 222 .5223 2.28222

 

.22222222 22.6 222 2.2222222: 22.2.... .2222.
22.2.2222... 2222. 2222-. 22 2222222. .222.-2 who

Sm 2..—ah.

68

 

 

 

222 222 222 222 222 222 222 222 lung
22. 22. 22. 22. .2. 22. 22. 22. 22.
8.2.22.2. 8.2.22.2. .22.2.22.2. .22.2.22.2 22.2.22... .2.2

..2.2.22. 822.22..

822.22.. 82.2.22.

.22.2.22. .22.2.22.

.22.2.22. 82.2.22.

22288.5 I:
22.22.2222 -...

 

2.22 2.22 2.22 2.22 2.22. 2.22. 2.22 2.22 22. 2222. 22 2222
.22.22 .22.22 .22.22 .22.22 .22.22 .22.22 .22.22 .22.22
22.22. 22.22 22.222 22.22 22.222 22.22. .2.22 .2.2. 2222228
. . ..22 .22.22 .22.22 .22.22
2. .22- 22.22.- 22.222- 22.22- 2.22.2.2 22
.2. .22 .22.22 .22.22 .22.22
2. . .2- 22.22.- 22.222- 22.22- 2.22.22 ...2
.22.22 .22.22 .22.22 .22.22
22.22- 22.22.- 22.22.- 22.22- 2.22.2.2 ...2
.22.22 .2..22 .22.22 .22.22
22.22- .222.- .2.22. - 22.22- 2.22.2.2 2.2
”oflm «OE

.22.22 .22.22 .2 . .22 .22.22 .22.22 .22.22 .2. .22 .22. .2 .222....22 2.2.2.. .2
.2.22- 22.22- 22.22- 22.22- 22.2. .- 22.22- 22.22- .2.2 . - 222.222
.22.22 ..2.22 .22.22 .22. .2 .22.22 .22.22 .22.22 .22.22
22.22- 22.22- 22.22- 22.22- 22.22.- .2.22- .2.22 22.22 2222222

.22 ..2 .22 ..2 .22 . .2 .22 ..2

22822: 202. 288: ..2. 2.8222: .22. 2.2.8: .02.
2.22222; .2825. ..2.22-— EEO .252... 2.222 20.22-— 22:22.2

 

22.222-«H 22.—22:35 2.2.2....— Esswuem .252 22.—22$:—
222222525 2...:— heaaw— .22 2822:2222.— 222222m— can: mac

Nm .2.2—ah-

69

 

 

222 222 222 222 222 222 222 222 22.2.2233;
22. 22. 22. 22. 22. 22. 22. .2. 22.
.22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 22.22222. .2.2.
.22.2.22.2 .22.2.22.2 .22.2.2. .2 82.2.2.2 82.2.2.2 .22.2.22.2 82.22.22 .22.2.22.2 22.25.22. ..2.
22.22.2222 -...
2.22 2.22 2.22 2.22 2.22. 2.22. ..22 ..22 2.2. 2222. 22 222:.
.22.22 .22.22 .22. .2 .22.22 .22.22 ..2. .2 .22.22 .22.22
22.22. 22.22 22.222 22.22 22.22 22.22. 22.22 . ..2 222.228
.22. .2 .22.22 .22.22 .22.22 .22. .2 .22.22 .. ...2 .22.22 .22.2.22.2 2.222. .2
22.2- 22.2- 2.2.- 22.2 22.22- .2.2- 22.. .- ...2- 2222222
.22.22 .22.22 ..2.22 .22.22
22.22.. 22. . 2.- 22.2. .- 22.22- 2.22.2.2 22
..2.22 .22.22 .22.22 .22.22
22.22- 22.22.- 22.22- 22.22- 2.22.2.2 22
.22.22 .22.22 .22.22 .22.22
22.22- 22.22- 22.22- 22.22- 2.22.2.2 2.2
.22.: .22.22 .22.22 .2..22
22.2.- 22.22- 22.22- 22.22- 2.22.22 2.2
nonm HOE
.22.22 .2..22 .22.22 .22.: .22.22 .22.22 .22.22 .22.22
22.22- 22.2.- .2.22- 22.22- 22.2. .- 22.22- 22.22 2..22 222220
.22 ..2 .22 ..2 .22 ..2 .22 ..2
2.2222: .2.2 8.2222: .22. 22822: .2.. 2822... 20.2
cocoa:- _2E_=< .5924 220 232-. 2.22 2033 228222

 

Augean-«H 2.22—22:3: 2.2—om...— 2:53.733

magmas-0:: «SQ—z ban—61— .«é nowaEmumm— meOOF—m— await-m WHO

mm 232,—.

70

 

.5

 

 

222. 222. 222. 222. 222. 222. 222. 222. 222. 22.322.238.22:
22. 22. 22. 22. 22. 22. 22. 22. 22. 22.
.22.2.22.2. ..2.2. 22.2 22.2 .22.2.22.22 22.2.22... .2.2
.22.2.22.2 .22.2.22.2 822.22.. 822.22.. .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.2. .2 22222.22 22.8822. ....
22.222225 -2.
2.. 2.. .2. .2. 22. 22. 2.22 2.22 22. 22: .22... 22 2222,.
.22.22 .22.22 .22.22 .22.22 .2..22 ..2.22 .22.: .22.22 .22.22
2. .2 2. . .2 22.2- 22.2 22.2- 22.2- 22.22 22.2- 2. .2 222.2220
.2..22 .22.22 .22.22 .22.22
22.2 22.2 22.2 22.22- 2.22.2.2 22
.22.22 .22.22 .22.22 .22.22
2.2 22.2 22.2 22.22- 2.22.2.2 22
.22.22 .. 2.2 .22.: .22.22
22.2 .2.2 22.2 22.2- 2.22.2.2 2.2
.22.: .22.: .22.22 ..2.22
22.2 .2.2 22.2 22.22- 2.22.2.2 ..22
UN? HOE
.22.22 .2..22 .22.22 .22.: .22.: .22.22 .22.: .22.: .22.22 .2328... 2.22... .2
22.2- . ..2- 22.2- .2.- 22.2- 22.- .2.2- .2.2 . ..2- 222.222
.22.22 .22.22 ..2.22 .22.: .22.: .22.22 .22.22 .22.22 .22.22
2. .2- 22.2- 22.2 .2.2- 22.2- 22.2- 2. .2.- 2.2- ...2- 222220
.22 . .2 .22 ..2 .22 . .2 .22 . .2 ..2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .22.22: .22 22.2
222952. 32.—Eon 23:22 ovum 22E

 

Ante-um— .uoaU use Begonia-w 223-2— =<Iawa~

20:22:32: 2..—:— 22a22—éez 2o moaafiamm 2.22-cm .232:— mAO
vm «Ea-F

71

 

 

22 82 MS 82 22 82 2.22 82 82 22.2.2. 22 222.622
22. .22. 22. .22. 22. 3.. ..2. .22.. .2. 222
7.2.2.2 . .. .22.2.22.2 .22.2.22.2 .22.2.22.2. 22.2.2222. 22.2.
..2.22. 22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 8.2.2.2.... 8.2.2.22... 8.2.2.22 82.25.. 8.2.2.8... 22.82.20 .2:
motmtaum u&
N.. 2.. ..2. ..2. :2. :2. 2.2. 2.2 2.2.. 22: .222. 22 2222
.2222 .2.2 .2..22 3222 .2.2 .2222 :32 .2222 .22.22
22.2 2.2 ..2.2- 8:..- 2.2.2 2.2.2- .22 2.2... .2.2 22222220
.222 ..22 .2222 ..2.22
2.2.2 8... 2.2... 2.2.2 . - 2.22.2.2 22
.22.22 .2222 .2...2 .822
8.2 .2.2 ....2 3... .- 2.22.2.2 .22
.22.22 .22.22 .2222 .22.22
2.2.2- .2.2 2.2.2 2.2.2.. 2.2.2.22 2.2.
.822 ......22 .22.22 .22.22
2.2.2. 2.2.2 8.2 22.- 2:22.22 em
HUN? “OE
.2..22 .22.22 . .2.2 .22.: .222 .22.22 .22.22 .222 .22.22 .222....22 2.222. .2
2.2- 2....- ..2.2- 8...- 2.2.2- 22.2. 22.2- 2. .- .2...2- 2222.52
.22.22 .22.22 .22.22 ..N..2 .822 .22.22 .22.22 .22.22 .22.22
2.2.2- 2.2. ..2.2- ..2.2- N. ..2- 2.2- 22.2- 3.2- 2.2.2. 222220
32 . .2 32 A .2 32 A .2 3.2 A .2 . .2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .22.22: 22.. 22.2
oEomwczm 23:50.2 2:52 voom 2.2.2.

 

Amuuobm 2.2.2933: Eaawuom .2.—w 852$:—
852—85 3...: ..2—«4-5272 .2.2 8:28:22— 8822”.— cofih 922
mm «3:.

72

 

 

 

2.2—gag:

 

2N2 R2 R2 R2 R2 R2 2N2
22. 22. 22.. 22. :2. 22. 22. 22.
8.22.2.8 .22.2.22.2 8.2.2. ..2.2. 222522.. 22.2.
82.222... ..2.2.... . 2... 8.2.2.8.. 8.2.2.22. 8.2.2.2.... 8.2.2.2.... 8.2.2.22. 22222.22. .2:
motmtwum -n—
22. 22. ..2. ..2. 2.22 2.22 m... 22: 2222. 22 222.2.
.22. .2 .22.22 .2.2 ..2. ..22 .22.22 .2...2 .22.22
22.2 m . .. 8.2 ..2.2- 22.2. 22.22 ..2 222.2220
.22.22 .N. ..22 .22.22
R... .2.2... 8.22- 2.22.2.2 22
..2.22 .822 .22.22
.222 ..22 .22. . m- 2.22.2.2 22
.22.22 ..2.2..2 .22.22
222 .2.2 22.22- 2.22.22 22
.22.22 .22.22 .22.22
2 . ..2 ..2... 2.2.22- 2.22.22 2...
UN? «OE
.2..22 .2..22 .222 .2.2 .232 ..2.22 .2022 .222252 2.22.2 .2
8.2- m . ..2- 8.2- 2 . ..2- 22.2- 2.2 22.2- 2222222
.22.22 .22.22 .22.22 .22.22 .22.22 ..N. .2 .22.. .2
8.2- 22.2- ..2.2- ..2.2- 22.22. 22.2- .2.....- 2222222
.22 ..2 .22 ..2 .22 ..2 ..2
:22: .z .2 :22: 22 .2 $288: 22. 33 $2885
oflewczm 23:22 223m 22< 2.2—2—

 

Amauotm 2.2—.0335 222,.— giant—lama

autumn—0:: «Ba—=— hODflJua—Gz kc mOuQEmumH such—H fiofimrfl mw—O

em «Eng.

73

APPENDIX C

74

 

 

 

 

 

Ila-.12: jg
.2. 22. 22. 22.
.22.2.22.2 .22.2.22.2
.22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2

j: . 2.2-112.522.232.32...
. 22. 22. 22. 22.

.22.2.22.2 22.22222. .2.2.

222.2: ..2 .22.2.22.2 22.222222. :22

22
22.2.2.2.
.22.2.22.2 .22.2.2. .2

22.22.2222 -2.

 

75

2.22 2.22 2.22 2.22 2.22. 2.22. 2.22 2.22 2.2. .222. 22 222.2.
.22.22 .22.22 .22. .2 .22.22 .22.22 .22.22 ..2.22 .22.22
22. .2 2..2.2 22.222 22.22. 22.22. 22.22 22.22 22. . 2 222222220
.22.22 .. 2.22 .2..22 .22.22
22.22- 22.22- 22.22- 22.22- 2.22.22 22
.22.22 .22.22 ..2.22 .2..22
22.2 . - 22.22- 22.22. 2.2. . 2- 2.22.2.2 22
.22.22 .22.22 .22.22 .22.22
22.2.- 22.2.- 2.22- 22.2- 2.22.22 22
.22.22 .22.22 .22.22 .22.22
22.2.- 22.2.- 22.22- 22.2.- 2.22.22 2..2

”0N? HOE

.22.22 .22.22 .22.22 .22.22 .22.22 .2 . .22 .22.22 .22.22 .222....22 2.222. .2
22. . .- 2.2- 22.2.- 22.. .- 22.22- .2.22- 22.2- 222. 222.222
.22.22 .22.22 .22.22 .2..22 .22.22 .2..22 .22.22 .2.2:
22.2.- 2.2. 22. .2- 2.... .- 22.22- 22.22- 22.22 22.22 222220

.22 ..2 .22 ..2 .22 ..2 .22 ..2

2228: 23 22222: 222 2322: ..22 258: 22.
22.22222 .2225 2222-. 2.2.0 2222-. 2.2.2 .223 2.22.22.

 

...-222222. 22.6 222 2.2222225 22.2.2. .2222.
22.222222..— 22222 222.3 22 2222822222 22222222 22...-... ago
8 2.222.

 

 

 

222 222 222 222 222 222 222 222 22.252.222.522:
22. 22. 22. 22. 22. 22. 22. 22. 22.
.22.2.22.2 22222.2 82.2.2.2 $22.22.. 22.52.22. 22.2.
822.22.. 2222.22. .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 22.52.22 .2:
22.22.2222 -...
2.22 2.22 2.22 2.22 2.22. 2.22. 2.22 2.22 2.2. 2222. 22 222.2.
.22.22 .22.22 ..2.22 .22.22 ..2.22 .22. .2 .2..22 .2..22
22.22 22.22 22. . 22 22.2.2 22.22. .2.22. 22.2.. 22.22. 2.222222
.22.22 .22.22 .2..22 .22.22
22.22- 22.22- 22.22- 22.2- 2.22.2.2 22
.22.22 .22. .2 .22.22 .22.22
22.22- .2.22- 22.22- 22. _- 2.22.22 22
.22.22 .2 . .22 .22. .2 .22.22
22.22- .2. .- 22. . 2- 22.2 2.22.22 22.
.22.22 .22.: .2..22 .22.22
22.22- 22.22- 22.22- .2.2- 2.22.2.2 22.2
”ON? “C:
.22.22 .22.22 .22.22 .22.22 .22.22 .22. .2 .2. .22 .22. .2 2222.222 2.222. .2
22.2.- 22.2.- 22.22- 22.22- 22.22. 22.22- .22.- 22.2.- 222.222
.22.22 ..2.22 .22.22 .22.22 .22.22 .22.22 .22. .2 .22.:
22.22- 22.2.- 22.22- 22.22- 22.22- 22.22- 222. 22.2. 222220
.22 ..2 .22 ..2 .22 ..2 .22 ..2
2822: 20.2 288: 222 288: 20.2 258: 22.
.22.22... .2832. 2.52-— EEU .5223 222 202.3 22:202.

 

Afluo-CH 39:02:25 2.22-2— E:.—w.8m 1:: 22:57.33

2222.22.22... 22...: 22.2-2 22 22.22.2222 2222.222. 22...-.— 2.5
20 2.2.22.

76

 

 

 

222 222 222 222 222 222 222 222 222. .2. ..2.22 222.22-I2
22. 22. 22. .2. 22. 22. .2. 22. 22.
.22.2.22.2 .22.2.22.2 .22.2.22.2 ..2.22222 22.222222. 22.2.
82.2.2.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 22222.22 .22.22. ..2 82.22.22 22.22.2222. 2.:
22.22.2222 -22
2.22 2.22 2.2 2. .2 2.22 2.22 2.22 2.22 222. 2222. 22 2.22.).
.22.22 .22.22 .22. .2 .22.: .2. .22 .22.22 .22.22 .22.22
2. 2.22 222.22 22.222 222.22 22.22 2 . 2.22 2.2.2.- 222.22- 222222220
.22.22 .22.22 .22.22 .22.22
.22.2.- 2222- 222.2 2222.- 2.22.2.2 22
.22.22 .. ...2 .22.22 .22.22
222.2- 2222.- 222.2 2222.- 2.2.2.22 22
.22.22 .22.22 .22.22 .22.:
222.2- 222.2- 222.2- 2222.- 2.22.2.2 2.2.
.22. .2 .22.: .22.22 .2. .22
222.2- 222.2- 222.2- 222.2- 2.22.2.2 222
”0N? “CE
.22.: .22.: .22.22 .2..22 .. ...2 .22.: .2..22 .22.22 .222....22 2.222. .2
.22.2- 222.2- 222. .- 222. .- 222.2- 22.2- 222.2- 22.. 222.222
.22.22 ..2.22 .2. .22 .22.22 .22.22 .2..22 .22.22 .22.22
222. .- 222.2- ..22 222.2 2.2.2.- 2222.- 222. . 2 222.22 222.220
.22 ..2 .22 ..2 .22 ..2 .22 ..2
288: Ba 22222: 20.2 288: 20a 2228: ..22
2.2.22..- _2E_:< 2.2.23 EEO 2.2223 222 2.53 22:22.,—

 

Amaoobm 3.252.512 2.2—0E 255327.53

magma—:5:— u-uA—fl— LAX-a2.— .ué muuasmumm navy—um taxman Mano

MU 22.2-r

77

 

 

 

22.. 22.. 22.. 22.. 22.. .22.. 22.. 22.. 22.. 22.2.2-32.222...qu
22. 22. .2. .2. 22. 22. 22. 22. 22. 22.
.22.2.22.2. ..2.2.22. .22.2.22.2 .22.2.22.2 22.62.22. 22.2.
82.222. 82.2.2.2. 82.2.22. .22.2.22. 82.2.22. .22.2.22. 82.2.2.2 .22.2.22.2 .22.2.22.2 22.6.55 :=
22.22.2222 -2.
22. ..2. 22. 22. 22. 22. 2.22 2.22 22. 222 .22... 22 222.).
.22.: .22.22 .22.22 .22.22 ..2.22 .22.22 .22.22 .2. .22 .22.22
22.2- 2.2- 22.2. 22.2- 222- 22.2- 22.2 22.... .2.2 2.22.2220
.22.22 .22.22 .22.22 ..2.22
2. .2 22.2 22.2 2.2.- 2.2522 22
.22.22 .22.: .22.22 .22.22
2.2 .2.2 22.2 22.2.- 2:22.22 22
.22.22 ..22 .22.22 ..2.22
2. .2 22.2 22.2 22.2.- 2.22.2.2 .2.2
.22.22 .22.: .22.22 .22.22
22.2 .2.2 22.2 22.2- 2.22.2.2 2.2
“on? HOE
.22.22 .22.22 .22.22 .22.: .22.22 .22.22 .22.22 .22.22 .22.22 .222....22 2.23.2. .2
22.2- 22.2- .2.2- .2.- ..2.2- ..2.2- 2. .2- 2.2. 22.2. 222.222
.22.22 .22.22 .22.22 .2..22 .22.22 .22.22 .22.22 .22.22 .22.2.2
2.2. 2.2. .2.2- 22.2- 22.2- 2. .2- 22.. .- 22.2- 22.2. 222.220
.22 ..2 .22 ..2 .22 ..2 .22 ..2 ..2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .22.2.22.2 222 22.2 .22.2.22.2
oEo_w:=-._ finite-.2 2:522 Bum 22< 2.2—2—

 

.2222222. 2220 222 2.222.222.2222 22.2.2. .2222.
22:22.2:— 2=._=— 2.5.5-2572 .2.2 2222:2222H 2.23.2.2”..— coxE mAO
22 2.2.22

78

 

 

 

222 222 222 222 222 222 222 222 222 flag-38:2:
22. 22. 22. 22. 22. .22. 22. 22. 22. 22.
.22..2. 2... .222. .2.. 822.22.. .22.2.22.2 22.5525 .2...
.22.2.22.2. 82:22.2. ..2.222 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 .22.2.22.2 22.222222. :22
22...-2.22%. -2.
2.. 2.. .2. .2. 22. 22. 22.2. 22.2. 2... 222 222.2. 22 22222,.
.22.22 .22.22 .22.22 .22.22 .22.22 .22.22 .22.22 .2..22 .22..2
22.2- 22.2 22.2- 22.2- 22.2- 22.2- 22.2.- 22.2.- 2.2 22222220
.22..2 .22..2 .22..2 .22.22
22.2 22.2 22.2 22.2. 2:22.22 22
.22..2 .22.22 .22..2 .22.22
22.2 .2.2 2.2 ..2.2. 2.22.2.2 22
.22..2 .22..2 .22..2 .22..2
22.2 22.2 22.2 ..2.22 2:22.22 22.
.22..2 .22.22 .22.22 .22.22
22.2 22.2 2.2.2 22.2. 2.22.2.2 .22
uonm HOE
.22.22 .22.22 .22.22 .22..2 .22.22 .22.22 .22.22 .22.22 .22.22 .222....22 2.222. .2
2. .2- 2.2- .2.2- .2.- 2. .2- ...2- 22. .- 22.2- 22.2 2222.222
.2...2 ..22. .2 .22.22 .22.22 .22.22 .22.22 ..2.22 .22.22 .22.22
22.2- 22.2- .2.2 22.2- 2.2. 2. .2- 22.2 ..2.2. 2.2.2- 222220
222 . .2 .22 ..2 222 . .2 .22 . .2 ..2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .22.2.22.2 22.. 22.2 22222225
22.2%...."— 2025222... 2:52: teem 22< 2.2—2—

 

Afioubm 3.5922822 222-.— 22.22 .252 Essa-8:32
8:22:22: 2..—:— ucaadécz .2.2 22afitmm 223.22% .22.:— mac

m0 03.5.

79

 

 

 

222 222 222 222 222 222 222 222 222 22.2.2 22 22.2822
22. 2... 22. 22. 22. .2. 22. 22. 22. 22.
.22.2.22.2 ..2.2.22. .22.2.22.2 .2..2222. 22.22225 22.2.
.22.2.22.2. .22.2.22.2 82.2.22. .22.2.22. 82.2.22. .22.2.22. 822.22.. 82.2.22. .22.2.22. 22.22.2222. I:
motmtfim on—
22. 22. .2. .2. 22. 22. .2..22 .2.22 22. 222 222:. 22 2222
..2.22 .22.22 .22.22 .2..22 ..2.22 .22.22 .22.22 .22.22 .22.22
2. .2- 2.2.2- 22.2 .2.2 22.2- 22.2- .2.22. 2.... . 22.. 222222220
.22.22 .22..2 .22.22 .2.:
22.2 22.2 22.2 22.... 2.22.2.2 22
.22.22 .22.22 .22..2 .22..2
22.2 22.2- 22.2 22.2. 2.22.2.2 22
..2.22 .22.22 .2.: ..2.22
22.2 .2.2- 22.2 22.2- 2.2.2.2.2 22
.22.22 .2. .22 .22.22 .22.22
2. .2 22.2- 22.2 22.2 2......2.2 .22
”ONE «CE
.2.2. .22..2 .22..2 .22.22 .22.22 .22..2 .22..2 .22222 .22.22 .22.-2.2222 2.25.2. .2
.2.2 222.2- .2.2 222. .2.2- 22.2- 22.2 ...2 22.2- 222.222
.22.22 .22.22 .22.22 .22. .2 .22.22 .22.22 .2.: .22.22 .222:
22.2- 22.2- 22.2 .2.2- .2.2- 22.2- 2.2.2. 2.2.2 22. .- 222220
.22 ..2 222 :2 .22 ..2 .22 a ..2
.2222 2. .2 .2222 2. .2 .2222 2. .2 .222222: 222 22.2 .2222222:2
2222522 ..2.2.—.2222. . 2:522 coom 22< BE

 

Eco-cm— Eosomacflv 222-.— 553.733
22:22:25 2:...— 2.5.5-52 Z .2.2 2225:3— 222.2-2.2m 1222 .2. mac
20 2..—ab

80

APPENDIX D

81

 

 

Table D1
h e r 89 1990

 

es 1
ion
Numerator Denominator
All fields Degrees of Degrees of
E Err-£12m 112229921 1217.311:
Ag Revenue
1) Gender, Year, HH dummies 3.29 159 2388 0.00
2) Gender, Year, Plot & HH dummies 3.36 163 2380 0.00 l 1
3) Gender, Year, Nonhead, Plot & HH dummies 3.35 163 2378 0.00
Female Labor
1) Gender, Year & HH dummies 1.29 159 2388 0.01
2) Gender, Year, Plot & HH dummies 1.40 163 2380 0.00 -
3) Gender, Year, Nonhead, Plot & HH dummies 1.26 163 2378 0.02
Male Labor
1) Gender, Year & HH dummies 1.62 159 2388 0.00
2) Gender, Year, Plot & HI-l dummies 1.79 163 2380 0.00
3) Gender, Year, Nonhead. Plot & HH dummies 1.82 163 2378 0.00
Child Labor
1) Gender, Year, HH dummies 1.82 163 2378 0.00
2) Gender, Year, Plot & HH dummies 2.25 163 2380 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 2.27 163 2378 0.00
Own Equipment Labor
1) Gender, Year & HH dummies 1.45 159 2388 0.00
2) Gender, Year, Plot & HH dummies 1.55 163 2380 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 1.61 163 2378 0.00
Fungicide Use
1) Gender, Year & HH dummies 2.16 159 2388 0.00
2) Gender, Year, Plot & I-IH dummies 2.29 163 2380 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 2.31 163 2378 0.00
82

 

Manure Use
1) Gender, Year & HH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Seed Use
1) Gender, Year & HH durmnies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Plot Size
1) Gender, Year & HI-I dummies

2) Gender, Year, Nonhead, Plot & HH dummies

83

0.78

0.87

159

163

163

159

163

163

163

159

2388

2380

2378

2388

2380

2378

2380

2386

0.06

0.09

0.07

0.00

0.00

0.00

0.98

0.87

 

 

Table D2
f h n 89 90

mm
1 1 iii i
Numerator Denominator
Millet and Sorghum Fields Degrees of Degrees of
E Enmlsun Enmlam Malia:

Ag Revenue

1) Gender, Year & 11H dummies 1.78 133 764 0.00
2) Gender, Year, Plot & HH dummies 1.94 137 746 0.00
3) Gender, Year, Nonhead. P101 & HH dummies 2.06 137 744 0.00
Ag Yield

1) Gender, Year & HH dummies 1.67 133 764 0.00
2) Gender, Year, Plot & HH dummies 1.83 137 746 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 1.97 137 744 0.00
Female Labor

1) Gender, Year & HH dummies 1.22 133 764 0.06
2) Gender, Year. P101 & HH dummies 1.25 137 746 0.04
3) Gender, Year, Nonhead. Plot & HH dummies 1.17 137 744 0.11
Male Labor

1) Gender, Year & HH dummies 1.52 133 764 0.00
2) Gender, Year, Plot & HH dummies 1.61 137 746 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 1.79 137 744 0.00
Child Labor

1) Gender, Year & HH dummies 1.08 133 764 0.27
2) Gender, Year, P101 & HH dummies 1.20 137 746 0.07
3) Gender, Year, Nonhead, Plot & I-IH dummies 1.38 137 744 0.00
Own Equipment Labor

1) Gender, Year & HH dummies 0.98 133 764 0.54
2) Gender, Year, Plot & HH dummies 0.95 137 746 0.64
3) Gender, Year, Nonhead, Plot & HH dummies 1.09 137 744 0.25

84

 

Fungicide Use
1) Gender, Year & HH dummies

2) Gender, Year, P1018: HH dummies
3) Gender, Year, Nonhead. Plot & HH dummies

Manure Use
1) Gender, Year & HH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Seed Use
1) Gender, Year & HH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, P101 & HH dummies

Plot Size
1) Gender, Year & HH dummies

2) Gender, Year, Nonhead, P101 & HH dummies

85

1.85

1.91

0.96

0.97

0.98

133

137

137

133

137

137

133

137

137

137

133

764

746

744

764

746

744

764

746

744

746

752

0.00

0.00

0.00

0.60

0.59

0.54

0.00

0.00

0.00

0.20

0.03

 

 

Table D3
W

W
M I I 5 '1] I'
Numerator Denominator

Peanut Fields Degrees of Degrees of

E Mam Freedom -V 1
Ag Revenue
1) Gender, Year & HH dummies 2.76 11 1 785 0.00
2) Gender, Year, Plot & HH dummies 2.82 115 777 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 2.91 115 775 0.00
Ag Yield
1) Gender, Year & HH dummies 3.53 111 785 0.00
2) Gender, Year, Plot & HH dummies 3.59 115 777 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 3.68 115 775 0.00
Female Labor
1) Gender, Year, HH dummies 1.67 11 1 785 0.00
2) Gender, Year, Plot & HH dummies 1.60 115 777 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 1.53 115 775 0.00
Male Labor
1) Gender, Year, HH dummies 1.76 111 785 0.00
2) Gender, Year, Plot & HH dummies 1.92 115 777 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 1.86 115 775 0.00
Child Labor
1) Gender, Year, HH dummies 0.58 111 785 1.00
2) Gender, Year, Plot & HH dummies 0.62 115 777 1.00
3) Gender, Year, Nonhead, Plot & HH dummies 0.60 115 775 1.00
Own Equipment Labor
1) Gender, Year & HH dummies 2.33 111 785 0.00
2) Gender, Year, Plot & HH dummies 2.42 115 777 0.00
3) Gender, Year, Nonhead, Plot & HH dummies 2.44 115 775 0.00

86

 

Fungicide Use
1) Gender, Year & I-IH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Manure Use
1) Gender, Year & HH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Seed Use
1) Gender, Year & HH dummies

2) Gender, Year, Plot & HH dummies
3) Gender, Year, Nonhead, Plot & HH dummies

Plot Size
1) Gender, Year & HH dummies

2) Gender, Year, Nonhead, Plot & 111-1 dummies

87

2.61

2.61

2.55

1.35

1.54

1.57

0.67

0.88

111

115

115

111

115

115

111

115

115

111

111

785

777

775

785

777

775

785

777

775

785

783

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.00

0.00

1.00

0.80

REFERENCES

Agarwal, Bina. “Gender and Command Over Property: A critical gap in Economic Analysis
and Policy in South Asia”. World Development, Vol 22, N010 (1994), 1010 -1046.

Alderman, Harold, Pierre-Andre Chiappori, et al. “Unitary Versus Collective Models of the
Household: Is it Time to Shift the Burden of Proof?” The World Bank Research
Observer vol 10, no. 1 (1995), 1-19.

Bardhan,P . “Size, Productivity and Returns to Scale: An Analysis of Farm-level Data in
Indian Agriculture”. Jouml of Political Economy, vol 81 (1973), 1370-1386.

Barrett, Christopher. “On Price Risk and the Inverse Farm Size-Productivity Relationship.”
Journal of Development Economics vol 51 (1996), 193-215.

Becker, Gary. A Treatise on the Fapnjy. Cambridge: Harvard University Press, 1998 (4th
edition).

. “A Theory of the Allocation of Time.” Economic Journal Vol 75

(1965), 493-517.

 

Ben-Porath, Yoram. “The F-Connection: Families, Friends, and Firms and the Organization
of Exchange”. Population and Development Review Vol 6 (1980), 1-30.

Benjamin, Dwayne. “Household Composition, Labor Markets and Labor Demand: Testing
for Separability in Agricultural Household Models, Econometrica Vol 60 (1992),
287 - 322.

. “Can Observed Land Quality Explain the Inverse Productivity
Relationship?” Journal of Development Economics, Vol 46 (1995), 51-84.

 

Browning, Martin, Francois Bourguignon, Pierre-Andre Chiappori, Valerie Lechene.
“Incomes and Outcomes: A Structural Model of Intrahousehold Resource
Allocation”. Journal of Political Economy, Vol 102, (1994), 1067- 1097.

Chiappori, Pierre-Andre. “Collective Labor Supply and Welfare”. Journal of Political
Economy, Vol 100, N03 (1992) 437 - 467.

 

. “Rational Household Labor Supply”. Econometrica 56 (1988), 63-90.

. Introducing Household Production in Collective Models of Labor Supply.
Journal of Political Economy, Vol 105, No 1 (1997), 191-209.

Collier, Paul. “Malfunctioning of African Rural Factor Markets: Theory and a Kenyan
Experience’, Oxford Bulletin of Economics and Statistics, V0145 (1983), 141 -1 72.

88

 

Dasgupta, Partha. An Inguig into Well-Being and Destitution. Oxford: Oxford University
Press, 1995.

Deaton, Angus. The Analysis of Household Surveys. Baltimore: John Hopkins University
Press, 1997.

de Janvry, Alain, Marcel Fachamps and Elisabeth Sadoulet. “Peasant Household Behavior
with Missing Markets: Some Paradoxes Explained”, The Economic Journal Vol 101
(November 1991), 1400- 1417.

 

Fachamps, Marcel. “Sequential Labor Decisions Under Uncertainty: An Estimable
Household Model of West African Farmers.” Econometrica. Vol 61, (1993), 1173-
1197.

Gebremedhin, Berhanu and Gerald Schwab. “ The Economic Importance of Crop Rotations
Systems: Evidence from the Literature” Michigan State University Staff Paper 98-13 ,
August 1998.

Gersovitz, Mark and John Waterbury, eds.: The Political Economy of Risk and Choice in
Senegal. London: Frank Cass, 1987.

Gladwin, Christina and Della McMillan. “Is a Turnaround in Africa Possible without
Helping African Women to Farm?” Economic Development and Cultural Change
Vol 37, no. 2 (1989), 345 - 369.

Goetz, Stephan. “Market Reforms, Food Security, and the Cash Crop - Food Crop Debate
in Senegal.” Unpublished Ph.D. Dissertation, MSU, East Lansing, Michigan, 1990.

. “A Selectivity Model of Household Food Marketing Behavior in Sub-
Saharan Africa.” American Journal of Agg'cultural Economics, Vol 74 (1992), 444 -
452.

 

Goldman, Noreen and Anne Pebley. “The Demography of Polygyny in Sub-Saharan Africa.”

In Reproduction and Social Organization in Sub-Saharan Africa, edited by Ron J.
Lesthaeghe. Berkely: University of California Press, 1989.

Guyer, Jane. “Food, Cocoa, and the Division of Labour by Sex in Two West African
Societies.” Comparative Studies Sociefl and Histog Vol 22 (1980), 355-73.

Haddad, Lawrence, John Hoddinot, and Harold Alderman. Intrahousehold Resource

Allocation in Developing Countries: Methods, Models, and Policy. Baltimore: Johns
Hopkins University, 1997.

Haddad, Lawrence and Thomas Reardon. “Gender Bias in the Allocation of Resources
within Households in Burkina Faso: A Disaggregated Outlay Equivalent Analysis.”

Journal of Development Studies, Vol 29 (1993), 260 -276.

89

 

Hopkins, Jane, Carol Levin, and Lawrence Haddad. “ Women’s Income and Household
Expenditure Patterns: Gender or Flow? Evidence from Niger.” American Journal
of Agg'cultural Economics Vol 76 (1994), 1219 -1225.

Jacoby, Hanan. “The Economics of Polygyny in Sub-Saharan Africa: Female Productivity
and the Demand for Wives in Cote d’Ivoire.” The Journal of Political Economy, Vol
103, no 5 (1995), 938-971.

Jones, Christine. “The Mobilization of Women’s Labor for Cash Crop Production: A Game
Theoretic Approach.” American Journal of Agricultural Economfi, Vol 65 (1983)
1049-1054.

Lele, Uma. “Women and Economic Transformation.” Economic Development and
Cultural Change. Vol 34 (1986), 195-221.

Lundberg, Shelly and Robert Pollack. “Separate Spheres Bargaining and the Marriage
Market.” Journal of Pplitical Economy Vol 101 (1993), 998-1010.

Kanbur, Ravi and Lawrence Haddad. “Are Better Off Households More Unequal or Less
Unequal?” Oxfprg Econonomip Pamg Vol 46 (1994), 445-458.

Kelly, Valerie, Bocar Diagana, et al. “Cash Crop and Foodgrain Productivity in Senegal:
Historical View, New Survey Evidence, and Policy Implications.” MSU
International Developing Working Papers, No. 20, 1996.

Kelly, Valerie, Thomas Reardon, et al. “Final Report for the IFPRI/ISRA Study of
Consumption and Supply Impacts of Agricultural Price Policies in the Peanut Basin
and Senegal Oriental.” Volume I, 1993.

McElroy, Marjorie and Mary Jean Homey. “Nash-Bargained Household Decisions: Toward
A Generalization of the Theory of Demand” International Economic Review, Vol 22,
(June 1981), 333-349.

Manser, Marilyn and Murray Brown. “Marriage and Household Decision-Making: A
Bargaining Anaysis”, International Ecgnomic Review, Vol 21, (1980), 31-45.

Reardon, Thomas. “Using Evidence of Household Income Diversification to Inform Study
of the Rural Nonfarm Labor Market in Africa.” World Development Vol 25, (1997),
735-747.

Reardon, Thomas and Edward Taylor. “Agroclimatic Shock, Income Inequality, and Poverty:
Evidence from Burkina F aso.” World Development Vol 24 (1996), 901-914.

. “Agroclimatic Heterogeneity, Income Diversification and
Inequality in Rural Burkina Faso.” Paper presented at the AAEA Meetings, Orlando,
Florida, 1993.

 

90

 

Reardon, Kelly, Crawford, Davadogo and Jayne. “Raising Farm Productivity in Afiica to
Sustain Long Term Food Security.” MSU Dept. of Agricultural Economics Staff
Paper, 94-77 (November 1994).

Reardon, Thomas, Christopher Delgado, and Peter Matlon. “Determinants and Effects of
Income Diversification Amongst Farm Households in Burkina Faso.” The Journal
of Development Studies. Vol 28 (1992), 264-296.

Rose, Elaina. “Ex Ante and Ex Post Labor Supply Responses to Risk in a Low Income
Area.” Unpublished Manuscript, revised 1995.

Rosenzweig, Mark R. “Neoclassical Theory and the Optimizing Peasant: an Econometric
Analysis of Market Family Labor Supply in a Developing Country.”

Quarterly Journal of Economics. Vol 94 (1980), 31-55.

Singh, Inderjit, Lyn Squire and John Strauss. Agg'cultural Household Models.
Baltimore: John Hopkins University Press, 1986.

Strauss, John. “Risk and Intrahousehold Analysis.” Paper Prepared for the External
Advisory Committee Meeting on Strengthening Development Policy through Gender
Analysis,” IFPRI, 18-19 March 1996.

Strauss, John and Kathleen Beegle. “Intra household Allocations: A Review of Theories,
Empirical Evidence and Policy Issues.” MSU International Developing Working
Papers, No. 62, 1996.

Udry, Christopher. “Gender, Agricultural Production, and the Theory of the Household.”
Journal of Political Economy. Vol. 4 , No 5 (1996), 1010 - 1045.

Udry, Christopher, John Hodinott, Harold Alerrnan, Lawrence Haddad. “Gender
Differentials in Farm Productivity: Implications for Household Efficiency and
Agricultural Policy.” Food Policy, V0120, No 5 (1995), 407-423.

. “Efficiency and Market Structure: Testing for Profit Maximization in Afiican
Agriculture”, Unpublished paper, 1998.

World Bank. “Rural Women in the Sahel and their Access to Agricultural Extension:
Overview of Five Country Studies.” September, 1994.

World Bank. “Senegal: Women in Development Country Assessment Strategy.” July 1992.

91

 

  
 

  

        

' 11111111.1111111'

3293