L: I; ... t.

 

4 , .
4T. .6. w...% _
”kw .....L..L
..

k... «3qu

. 2a: .

t! . .

.Mn. .. . 8km H.
.L .L Lang.

43.2.:

,L .
.q
an .3
fl: . ,
5v. _ L . .. Ln,
.1 l'rum . L . .‘omgfiétvvlfi .

_ flaw... L . 5.56

rfin . L

a: L
.49

I

.. .,:_ n.“
.1? EM

2??

. I on;
£41,. :0: p

I! {Gt-.1:
Vial 15....

as}! I,
3%... L

 

THESiS‘

IHII301"H!IIIWIHIIIIHIHill!WilliIHIHHIIUHIIIII

31293 01771 8382

LIBRARY

Michigan State
Unlverolty

 

 

 

 

This is to certify that the
dissertation entitled

INCREASES IN WOMEN'S EDUCATION AND FERTILITY DECLINE:
EVIDENCE FROM NIGERIA

presented by

Agatha Dadzie Awuah

has been accepted towards fulfillment
of the requirements for

PH.D degree in ECONOMICS

 

   
 

Ma' p fessor

Date 5 a 3 8

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

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

 

DATE DUE DATE DUE DATE DUE
_—"’6

()q‘ ‘
Let}: 2 b we
‘9‘

 

 

 

@?1301
Novfiozsjzoiid'5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1M chlRClDfiDu‘pGS-p.“

INCREASES IN WOMEN'S EDUCATION AND FERTILITY DECLINE: EVIDENCE
FROM NIGERIA

By

Agatha Dadzie Awuah

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Economics

1998

ABSTRACT

INCREASES IN WOMEN'S EDUCATION AND FERTILITY DECLINE : EVIDENCE
FROM NIGERIA

By

Agatha Dadzie Awuah

Fertility transition has been slow in sub-Saharan Africa. Recently, however,
there is evidence to suggest a fertility transition in some countries and some major
regions in sub-Saharan Africa. Notable among these is Nigeria where Cohen (1993)
asserts that there has been a drop of 1.3 in the total fertility rate. Reinis et a1 (1991) also
suggest a 10 percent drop in fertility among women in the Southwest and possibly the
Southeast.

Using Nigeria Demographic and Health Survey data, 1990, this study investigates
if fertility has indeed fallen and if so among which socio-economic groups. Using both
bivariate and multivariate techniques, the study starts out by looking at the trend in
female schooling over recent years. The data suggest a tremendous increases in female
schooling. Secondly, the study investigates the fertility behavior of women in the survey
as a whole and also with reference to the three major regions in Nigeria, the Southeast,
Southwest and the North. From the retrospective birth histories, we reconstructed
cumulative fertility at various attained ages. A look at fertility using such indicators as
cumulative fertility at various attained ages, first marriages and first births at various
attained ages does not reveal any decisive decline in fertility. Among the two youngest

cohorts there is some evidence of a decline in the Southwest and the Southeast for

cumulative fertility at ages 18 and 23 but there is no evidence of a decline at ages 28 and
beyond, for older cohorts.

Our regression analysis reveals a powerful negative relation between female
schooling and fertility among women who have completed primary and beyond. At
lower levels of schooling the relation is either positive or nonexistent. Our results also
suggest that female schooling alone accounts for between 23 to 61 percent of the decline
in fertility among the youngest cohorts. Among women of different socio-economic
status, it is women with seven or more years of schooling for whom we observe a
decisive decline in fertility. Fertility has in fact risen among women with either no
schooling or 1-6 years of schooling. For women with seven or more years of schooling

we also observe delays in the onset of first births and first marriages.

Copyright by
AGATHA DADZIE AWUAH
1998

This dissertation is dedicated to: My late father Papa Kofi F aasemkye., my mother
Maame Ama Ata and my husband, Emmanuel Baffour Awuah

ACKNOWLEDGMENTS

I would like to thank God for the strength he provided to carry me through to the
completion of my Ph.D. program in economics. People made themselves available to be
used to meet my need along the way and in the process I have become indebted to many.

First, I would like to express my sincere thanks and appreciation to Professor John
A. Strauss, my major advisor, for his guidance, support, cooperation, and assistance in the
direction and completion of my dissertation. I have been very fortunate to have him as
my advisor. I am also indebted to the other members of my dissertation committee,
Professors Nan Johnson, Carl Liedholm and Rowena Pecchenino for carefully reading the
manuscript and making many valuable suggestions and comments which helped shape
the final product.

My sincere thanks also to Dr. John Metzler, Dr. David Wiley and Dr. Jacob
Fisseha of the African Studies Center for providing me with research assistantship
position which helped financially my doctoral program. To the secretarial staff, JoAnne
Peterson, Lisa Beckum, Melissa Jeffrey and Terry McCaskey, thanks for the word
processing tips and cordial working relationship.

To all my high school teachers, Mr. Buhari, Mr. Ntiamoah, Mr. Amoah-Bediako
and Mr. Nditi Segbezi, who saw in me the promise of a college graduate and provided the
incentives for higher learning.

I am deeply indebted to Jack and Margaret Jones, both medical professors in the

School of Human Medicine here at Michigan State University, for their financial,

physical and unconditional love and support which made this a possibility. Your labor of
love will be rewarded.

To my mother, two brothers and two sisters back in Ghana for the sacrifice, pain
and financial hardship you endured during my long pursuit of knowledge. Thanks for
bearing with me.

Finally, to my dear husband Dr. Emmanuel Baffour Awuah whose sacrifice in
terms of love, time, patience, financial, physical, spiritual and emotional support made it
happen. To my two wonderful children, Eunice and Yaw Awuah, who were born and
grew up during my student years. Thanks for your understanding when my role as a
student ofien conflicted with my role as a mother and a wife. I honestly could not have

done it without your help.

vii

TABLE OF CONTENTS

 

 

 

 

 

 

 

LIST OF TABLES .............................................................................................................. xi
LIST OF FIGURES .......................................................................................................... xiv
CHAPTER I
INTRODUCTION ............................................................................................................... 1
Introduction to the Problem Setting ........................................................................ 1
Objectives of the Study ........................................................................................... 3
Significance of the Study ......................................................................................... 5
Organization of the Study ........................................................................................ 6
CHAPTER 2
BACKGROUND INFORMATION ON NIGERIA ............................................................ 8
History .......................................... 8
Geography .............................................................................................................. 10
Economy ................................................................................................................ l 1
Population .............................................................................................................. 12
Education .............................................................................................................. 13
Population Policies ................................................................................................ 14
CHAPTER 3 ~
REVIEW OF LITERATURE ............................................................................................ 17
Introduction - - ................................................................................. l 7
Women's Education and Fertility ........................................................................... 18
Research on Education and Fertility - ........ 21
Research on Nigeria - - .......................................... 32
Problems Associated With Empirical Work on Fertility ....................................... 36 -
CHAPTER 4
POTENTIAL MEASUREMENT ERRORS IN THE NIGERIA
DEMOGRAPHIC AND HEALTH SURVEY, 1990 ......................................................... 39
Introduction - ................................................................ 39
Data Collection ...................................................................................................... 41
Age Reporting ............................................................................................ 41
Birth Histories ............................................................................................ 43
Age at First Marriage ................................................................................. 43
Age at First Birth ....................................................................................... 45
Measurement Errors -- ....................................................................... 46
Age Heaping .............................................................................................. 46
Respondents Age ........................................................................... 46
Age at First Marriage and Age at First Birth ................................. 48

viii

Incentive for Misreporting by Enumerators ........................................................... 52

Underreporting due to Memory Lapse by Respondents ........................................ 53
Completeness of Information on Date of Birth
of Respondents ........................................................................................... 54
Completeness of Information on Birth Histories
of Respondents ........................................................................................... 59
Proportion of Deaths .................................................................................. 6O
Cumulative Fertility at Various Attained Ages ......................................... 63
Cumulative Fertility at Various Attained Ages by
Years off Schooling .................................................................................. 67
Is There a Baby Boom Effect? .................................................................. 72
Summary and Conclusions .................................................................................... 77
CHAPTER 5
FERTILITY TRENDS IN NIGERIA ............................................................................... 80
Introduction ........................................................................................................... 80
Trends in Female Schooling .................................................................................. 82
Fertility Trends in Nigeria ..................................................................................... 87
Number of Children Ever Born ................................................................. 87
Cumulative Fertility: Another Look .......................................................... 90
Birth Patterns ............................................................................................. 95
Marital Patterns .......................................................................................... 98
Conclusion ............................................................................................... 102
Family Planning Policies and Contraceptive Use ................................................ 103
Family Planning Policies ......................................................................... 103
Organization and Availability of Family Planning
Services .................................................................................................... 103
Contraceptive Use .................................................................................... 104
Summary and Conclusions .................................................................................. 107
CHAPTER 6
THE DYNAMIC RELATIONSHIP BETWEEN FEMALE SCHOOLING
AND FERTILITY ........................................................................................................... 110
' Introduction ............ '. ............................................................................................. 110
The Theoretical Model ......................................................................................... 112
Empirical Specification of the Model .................................................................. 115
Omitted Variables and Measurement Error Problems ............................. 118
Omitted Variable .......................................................................... 1 18
Measurement Error ...................................................................... 119
Methodology ........................................................................................................ 120
Measurement of Variables ....................................................................... 120
Methodology ............................................................................................ l 24
Empirical Results ................................................................................................. 124
Education and Fertility: Mean CEB and Completed
Years of Schooling ................................................................................... 125

ix

Regression Analysis -- .............................................. 128

 

Impact of Education on Fertility

Female Schooling only ................................................... 128

Female Schooling and Age Cohorts ............................... 130

Female Schooling, Age Cohorts and

Community Variables ...................................................... 133

Predictions: The Decline in Fertility Accounted

for by Female Schooling ................................................. 135

Age at first marriage and Age at first Birth ..................... 136
Age at First Marriage ........................................... 136
Age at First Birth ................................................. 137

Female and Husband's Education with Cohort
and Community Variables

 

 

 

(Ever-Married Women) ................................................... 141
Urban and Rural Samples ................................................ 141
Summary, Conclusions and Implications ............................................................ 150
CHAPTER 7
INCREASES IN WOMEN'S EDUCATION AND FERTILITY
DECLINE ....................................................................................................................... 153
Introduction ..................................................................................... l 53
Increases In Women's Education and Fertility Decline ...................................... 154
Cumulative Fertility at Various Attained Ages
by Years of Schooling ........................................................................... 154
Education and Fertility: Results by Levels
of Female Schooling ................................................................................ 157
Age at First marriage and Age at First Birth:
By levels of Female Schooling ................................................................ 160
Age at First Marriage ................................................................... 160
Age at First Birth ......................................................................... 162
Summary - - -- ..................... 164
CHAPTER 8
SUMMARY, CONCLUSION AND IMPLICATIONS .................................................. 174
Introduction ........................................ - - ......................................... 174
Summary .............................................................................................................. 174
Findings and Conclusions .................................................................................... 179
Implications and Suggestions for Further Research ............................................ 180
APPENDIX .................................................................................................................... 184
REFERENCES ................................................................................................................ 195

LIST OF TABLES

Table 4.1-Distribution of Age of the Respondents, Age at First Maniage
and Age at First Birth ................................................................................. 51

Table 4.2-Percent Distribution of All Women by Completeness of information on
Date of Birth, by Age group and by Years of Schooling. Nigeria
Demographic and health Survey. 1990 .................................................... 57

Table 4.3-Mean Number of Deaths per 1,000 Live Births by Years of
Schooling, Nigeria, 1990 ..... ........................................... 62

 

Table 4.4-Cumulative fertility At Various Attained Ages By Five-Year Age group.
Nigeria, 1990 .............................................................................................. 66

Table 4.5-Cumulative Fertility at Various Attained Ages By Five-Year
Age Group and by Years of Schooling, Nigeria, 1990 ............................. 71

Table 4.6-Mean Number of Children Ever Born Three Years Prior,
During and After the Nigerian Civil War ................................................ 76

Table 5.1-Distribution of All Women by Number of Children Ever Born (CEB)
and Mean Number Living Children by Five-Year Age group.
Nigeria, 1990 .............................................................................................. 88

Table 5.2-Percent of Women with First Births at Various Attained
Ages By Five-year Age Group. Nigeria, 1990 ......................................... 97

Table 5.3-Percent of Women Ever-Married at various Attained
Ages By Five-Year Age Group. Nigeria, 1990 ....................................... 101

Table 6.1-Impact of Education on Fertility-Female -
Female Schooling only ............................................................................ 131

Table 6.2-Impact of Education on Fertility - Female Schooling
~ and Cohort Effects................... ................................................................ 132

Table 6.3- Impact of Education on fertility - Female Schooling,
Cohort Effects and Community Variables .............................................. 134

Table 6.4-Impact of Education on First Marriages at Various Attained Ages -

Female Schooling, Cohort Effects and
Community variables (Logit) ................................................................... 139

xi

Table 6.5- Impact of Education First Births at various Attained Ages -
Female Schooling, Cohort Effect and
Community Variables (Logit) ................................ 140

 

Table 6.6-Impact of Education on Fertility - Female Schooling,
Cohort Effects, Community Variables and Husband's
Education (Ever-married at Age 18) ....................................................... 144

Table 6.7-Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education
(Ever-married at Age 23) ......................................................................... 145

Table 6.8-Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education
(Ever-married at Age 28) ......................................................................... 146

Table 6.9-Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education
(Ever-married at Age 33) ......................................................................... 147

Table 6.10-Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education
(Ever-married at Age 38) ......................................................................... 148

Table 7.1-Impact of Education on Fertility - Female Schooling, Cohort
Effects and Community Variables by levels of
Female Schooling .................................................................................... 167

Table 7.2-Impact of Education on Fertility - Female Schooling, Cohort Effects and
Community Variables by levels of
Female Schooling ................................................................................... 168

Table 7.3-Impact of Education on Fertility - Female Schooling, Cohort Effects and
Community Variables by levels of
Female Schooling ..................................................................................... 169

Table 7.4-Impact of Education on Age at First Marriage - Female Schooling,
Cohort Effects and Community Variables by levels of
Female Schooling (Logit) ...................................................................... 170

Table 7.5-Impact of Education on Age at First Marriage - Female Schooling,

Cohort Effects and Community Variables by levels of
Female Schooling (Logit) ...................................................................... 171

xii

Table 7.6-Impact of Education on Age at First Birth - Female Schooling,
Cohort Effects and Community Variables by levels of
Female Schooling (Logit) ........................................................................ 172

Table 7.7-Impact of Education on Age at First Birth - Female Schooling,
Cohort Effects and Community Variables by levels of
Female Schooling (Logit) ....................................................................... 173

Table A5. l-Distribution of Women by Years of Schooling, Female
Cohorts Born 1943-72. Nigeria, 1990 ................................................... 185

Table A5.2-Percentage of All Women with Second and Third or more
Births at Various Attained Ages .............................................................. 187

Table A6.1- Prediction of the Decline in Fertility Attributable to
Female Schooling: Cumulative Fertility
at Age 18 ................................................................................................. 188

Table A6.2- Predictions of the Decline in Fertility Attributable to
Female Schooling: Cumulative fertility
at Age 23 ................................................................................................ 189

Table A6.3-Impact of Education on Fertility - Female Schooling,
Cohort Effects and Community Variables
(RURAL RESIDENCE) ......................................................................... 190

Table A6. 4-Impact of Education on F ertrlrty Female Schooling,
Cohort Effects and Community Variables
(URBAN RESIDENCE) - - .......................................... 191

 

Table A6.5-Impact of Education on Fertility - Female Schooling :
Results by Age Cohort ............................................................................ 192

Table A6.6-Impact of Education on Fertility - Female Schooling and
Cohort Effects: Results by Age Cohort ................................................... 193

Table A6.7-Impact of Education on Fertility - Female Schooling,

Cohort Effects and Community variables:
Results by Age Cohort ............................................................................ 194

xiii

LIST OF FIGURES

 

Figure 2.1- Map of Nigeria--- -- .................................................................... 16
Figure 4.1-Age Distribution of Respondents ..................................................................... 50
Figure 5.1A-Mean Years of Schooling. Female Cohorts Born 1943-72 .......................... 84
Figure 5.1B-Percentage Distribution of Years of Schooling.

Female Cohorts Born 1943-72 ................................................................... 84
Figure 5.1C & 5.1D- Mean Years of Schooling.

Female Cohorts Born 1943-72 .................................................................. 85
Figure 5.1E & 5.1F- Percentage of Women with Zero Years of Schooling.

Female Cohorts Born 1943-72 ................................................................. 85
Figure 5.1G-Percentage of Women with 6+ Years of Schooling.

Female cohorts Born 1943-72 ................................................................. 86
Figure 5.1H-Percentage of Women with 6+ Years of Schooling.

Female Cohorts Born 1943-72 ................................................................. 86
Figure 5.2A-Cumulative Fertility at various Attained Ages.

All Women, Nigeria, 1990 ......................................................................... 94
Figure 5.2B-Cumulative Fertility at Various Attained Ages .

Southeast, Nigeria, 1990 ............................................................................ 94
Figure 5.2C-Cumulative fertility at various Attained Ages.

Southwest, Nigeria, 1990 ........................................................................... 94
Figure 5.2D-Cumulative Fertility at various Attained Ages.

North, Nigeria, 1990 .................................................................................. 94
Figure 6.1-Mean CEB by Completed years of Schooling.

Female Cohorts Born 1943-72 ................................................................ 127
Figure 7.1A-Cumulative Fertility at Age 18 by Years of Schooling ............................... 165
Figure 7.lB-Cumulative fertility at Age 23 by Years of Schooling ................................ 165
Figure 7.lC-Cumulative Fertility at Age 28 by years of Schooling ................................ 165

xiv

Figure 7.lD-Cumulative fertility at Age 33 by Years of Schooling ................................ 165 A

Figure 7.2A-Cumulative fertility at various Attained Ages
by Years of Schooling (0 years) ............................................................... 166

Figure 7.2B-Cumulative fertility at various Attained Ages
by Years of Schooling (1-5 years) ........................................................... 166

Figure 7.2C-Cumulative fertility at various Attained Ages
by Years of Schooling (6 years) .............................................................. 166A

Figure 7.2D-Cumulative fertility at various Attained Ages
by Years of Schooling (7+ years) ........................................................... 166A

XV

 

 

CHAPTER 1
INTRODUCTION

1.1. INTRODUCTION TO THE PROBLEM SETTING
There is a close relationship between the population growth rate and economic
growth and development even though how population issues should be handled is
subject to debate amongst development economists. One of the arguments for a
reduction in the growth rate of population is that a reduction is good in itself because it
raises the per capita Gross National Income (GNP). Those who argue for a natural
growth rate in population maintain that labor is a key ingredient in the production
process. More could be achieved for every one, if the polluters of the environment are
made to clean the environment or degradation of forest could be restored if new trees
are planted. A nation’s population determines among other things the potential labor
supply and the labor force participation rate. However, the quality of the labor force as
measured by the level of education and technical training is most important to any
economic growth process. In spite of these controversies, it is clear that regions with

high population growth rate and high total fertility rates are also the poorest in the

World.
Sub-Saharan Africa seems to lag behind the fertility transition that has begun in

eVery other major region in the developing world since 1965. For example, by 1997
figures, the population growth rate between 1990-95 for sub-Saharan Africa was 2.6

percent. This compares with 1.6 percent average growth rate for all low and middle

income countries, 0.3 percent for Europe and Central Asia, 1.7 percent for Latin
America and Caribbean and 0.7 percent for high-income countries. On the other hand
about 31 out of the 49 low-income countries are sub-Saharan African countries. Only
two out of the 41 countries classified as middle income countries are from sub-Saharan
Africa. South Africa and Gabon are the only sub-Saharan Africa countries that fall
within the upper-middle income category. Most sub-Saharan African countries had a
negative growth rate in GNP for the period between 1985-95. Sub-Saharan Africa as
a region recorded -1.1 percent growth rate, in comparison, that of all low-income
countries was 0.4 percent, Latin America and Caribbean was 0.3 percent and the
United States 1.3 percent (World Development Report, 1997).
Population issues seem to be a major factor in sub-Saharan Africa’s
development (Caldwell, 1995). Many reasons are given for Africa’s resistance to a
fertility transition. Key amongst these reasons are unique cultural and religious
practices in the continent which promote the demand for larger families (Caldwell and
Caldwell, 1987, 1990). Some also argue that Africa lags behind because her level of
development is not at par with that of other regions where fertility transition has taken

place (World Bank, 1984, 1986).

However, in spite of these problems many researchers have found that even in
hi gh fertility regions like sub-Saharan Africa, fertility differentials exist among different
SOcio-economic groups (Cochrane and Farid, 1990). Botswana, Zimbabwe and Kenya

are often cited as the vanguard of the fertility transition in sub-Saharan Africa. Recently

however, from Cohen (1993) extensive fertility analysis of African countries, it has

become apparent that fertility transition is underway in other African countries besides
Botswana, Kenya and Zimbabwe, and some major regions in Africa. Notable among
these countries is Nigeria. By 1991, there was evidence to suggest a 10 percent decline
in fertility in the southwest and possibly the southeast Nigeria (Reinis, Rutstein and
Ajayi, 1991). The claim of a fertility decline in Africa is however, met with skepticism
because of problems associated with data collection in the region. Moreover, by 1985
World Fertility Survey (WFS) on Nigeria, her total fertility rate was 5.7 and according to
1997 figures this has risen to 6.2, which represents an increase but not a decline in
fertility (Scott and Chidambaram, 1985; Population Reference Bureau, 1997).
The present study focuses on Nigeria. Our aim is to track fertility in Nigeria to
see if it has indeed fallen and if so among which socio-economic group. Verification of
this recent drop in fertility in the southwest and southeast subregions in Nigeria is of
interest because by 1991 census data, southwest Nigeria was over 36 million, which is
more populous than any sub-Saharan country outside Nigeria, except Ethiopia, South
Africa and Zaire. In addition southwest and southeast Nigeria together is larger than any

sub—Saharan country in Africa.

#2. OBJECTIVES OF THE STUDY
Using the Nigeria Demographic and Health Survey (NDHS), 1990 data, we have
three objectives in this study.
* First to track fertility trends in Nigeria to see if it has indeed fallen, by

e)Iiamining the fertility behavior in Nigeria using key correlates of fertility such as

cumulative fertility, age at first marriage and age at first birth. A sustainable fertility
decline must be accompanied by younger women marrying later and having their first
births later as compared to the older women. The usual practice of comparing the fertility
of women at two pints in time, for instance, eight and four years before a survey, may be
an inaccurate measure of a fertility decline. Our objective in this study is to do a cohort
analysis to see how fertility patterns are changing from the oldest to the youngest women
in the survey. To investigate if younger women consistently differ in their fertility
behavior from the older women.

* Second, if fertility has indeed fallen, we would like to investigate further among
which socio-economic group fertility has fallen using female schooling as our measure of
social stratification. This we hope to accomplish by doing fertility analysis for women

with different levels of education.

* Third, to find out how much of the decline in fertility is accounted for by female

education. Using Oaxaca (1973) decomposition as applied to male-female wage
discrimination, we will decompose the decline in fertility if any, into that which is
eXplained by female schooling and hence by other factors besides female schooling
- which may affect fertility exogenously.
We begin by examining the consistency and reliability of the data. We also
iIlil‘restigate the trend in female schooling to find out if younger women have received
hi gher education relative to the older women. If so, whether the increases in female

Schooling may: account for the any possible decline in fertility.

1.3. SIGNIFICANCE OF THE STUDY

First the study provides an alternative to the use of children ever born (CEB) as
the dependent variable in regressions that seek to find the determinants of fertility. There
is a problem with this approach that the present study seeks to address. We use
cumulative fertility at various attained ages as our dependent variable.

There are immediate advantages to the new approach. It avoids censoring
problems and the age at which the relationship between female schooling and fertility
diminishes can easily be captured even without multivariate regression analysis.
Research in this line has not been done for an African country even though Lam,
Sedlacek and Duryea (1992) have done this using Brazilian data. However, doing it for

an African country such as Nigeria is of interest in itself. With a population which is 14
percent of the Africa’s total population and a population grth rate of 3.1 percent,
Nigeria offers herself as a good starting point for research which has a substantial bearing
on Africa’s population future.

It is also the first of its kind which offers a systematic quantification of female
svclrooling on fertility as applied to an African country. If policy makers are to put some
emphasis on female schooling as a vehicle for a reduction in fertility, then it is important
to know how much an increase in female schooling will lead to a decline in fertility. It is
211 so important to know the age in a woman’s life-cycle an intervention in her education is
l"lecessary to yield maximum results on her fertility decisions.

The study will also contribute in general to the theoretical, empirical and policy

discussions on the relationship between maternal education and fertility. Studies on

fertility and education are many but none or very few seek to quantify the influence of
female schooling on fertility. Knowing how much of a decline in fertility is accounted
for by female schooling alone is significant for policy making which seeks to put some
emphasis on female schooling. This is beneficial for Africa in general where private
investment in girls' education is low. The study will also benefit the international
community that seeks to advance the potentials of women through education and

reproductive health.

1.4. ORGANIZATION OF THE STUDY

 

The rest of the study is organized as follows: Chapter 2 presents a brief
background information on Nigeria, making reference to its history, geography,
economy, population, education, and population policies. Chapter 3 reviews some of the
relevant theoretical and empirical literature, with particular reference to some of the
stylized facts on the relationship between female schooling and fertility. It also discusses
some of the empirical work done from both the World Fertility Surveys and
Demographic and Health Surveys on Africa and other parts of the world. Finally it
di scusses some of the empirical work on Nigeria.
Chapter 4 investigates some of the potential measurement errors in the NDHS,
1 990 data, with particular emphasis on data collection on birth histories, age at first
t1larriage and age at first birth, and completeness of information on age of respondents,

811d the proportion of dead live-births. Chapter 5 provides the descriptive statistics on the

trends in female schooling and fertility trends in Nigeria with reference to cumulative
fertility at various attained ages, first marriage and first births at various attained ages.
Chapter 6 deals with multivariate regression analysis capturing the dynamic
influence of female schooling, community variables such as urban residence and religious
background, polygyny and husband’s education on fertility, age at first marriage and age
at first birth. Chapter 7 investigates further among which socio-economic group fertility
may have indeed fallen by running regressions over levels of female schooling for both
cumulative fertility at various attained ages, first marriage and first birth at various

attained ages. Chapter 8 provides the summary, conclusions and implications and

suggestions for further research.

CHAPTER 2
BACKGROUND INFORMATION ON NIGERIA

2.1. HISTORY
The history of the present day Nigeria dates largely from 1914 when the

previously distinct northern and southern Nigerian protectorates were amalgamated under
British rule. Nevertheless, oral traditions of the various ethnic groups, documentary
sources and archaeological evidence reveal the long existence of various dynamic and
interacting kingdoms within its boundaries thousands of years before the influence of the
Europeans. Documentation exists for example on the iron-making culture such as the
Nok who were living on the Jos Plateau as early as the 300 BC. Other important
kingdoms were the northeastern kingdoms of Bomo, the Hausa city-state kingdoms of
Katsina, Kano, Zaria, and Gobir in northem-central Nigeria, the Yoruba city-state
kingdom of Ife, Oyo and Ijebu in southwestern Nigeria, the southern kingdom of Benin

and the Igbo communities of eastern Nigeria (Europa Publications, 1996: NewAfiica

Yearbook, 1997-98; Oyewole, 1987).
Contacts between the peoples of Nigeria and Europeans date at least to 1480 when

two Portuguese ships explored along the Bight of Benin and bought slaves. Soon
Elnopean powers were exchanging cloth, hardware, guns and gunpowder for slaves along

the West African Coast. The abolition of slavery by the British in 1807 cleared the way

for the expansion of trade in agricultural produce from Africa to Europe.

At the end of the 19th century, Britain began aggressive military expansion in the
region. A protectorate was declared over Lagos, other parts of the south and in the
northern areas from 1885-1900. The North and the South were formally consolidated in

1914. In the North, the British limited Christian missions, restricted education, and
reinforced the feudal rulers. In 1939, Eastern and Western Nigeria were separated
leading to the structure of three separate regions which was in place at the time of

independence in 1960. Within each region, one ethnic group predominated, the Hausa-

Fulani in the North, the Yoruba in the Southwest and the Igbo in the Southeast.
On October 1, 1960, Nigeria became an independent nation. Three years later in
October 1963, it became the Federal Republic of Nigeria, severing all links with the
British, while retaining its membership in the Commonwealth. The political scene
leading up to independence was dominated by three regionally based parties, the National
Council for Nigeria and Cameroon (N CNC) in the east, the Action Group (AG) in the
West, and the conservative Northern People’s Congress (NPC) in the north. Independent
N i geria was almost immediately confronted by a series of foreseeable problems.
A complex series of political events, constitutional difficulties, ethnic allegiances
atld social unrest followed by corrupt elections in 1964 and 1965, paved the way for the
f‘llt‘st of six unplanned changes of government which have occurred since Nigeria became
independent. In 1966, the middle-ranking members of the Nigerian military staged an
attempted coup. This was suppressed by federal troops, but resulted in the installation of
a military junta, led by Igbo officers. Regional animosities flared, promoting massacres

0f Igbo-speaking people living in the north. The following year, eastern leaders

responded by declaring a separate Republic of Biafra, igniting a 30-month civil war from
May 1967 to January 1970, which claimed two million lives. Since 1960 Nigeria has

been run by a civilian government by slightly less than 10 years (Europa Publications,

1996: NewAfrica Yearbook, 1997-98; Oyewole, 1987).

2.2 GEOGRAPHY

The Federal Republic of Nigeria is located in West Africa surrounded by Niger on
the North, Cameroon on the East, and Benin on the West. It covers an area of about
923,768 square kilometers, an area roughly twice the size of California and nearly four

times that of the United Kingdom. It has about three percent of Afiica’s land surface but
nearly 14.1 percent of its population, making her by far the most populous country in
Afi-ica. Figure 2.1 is a map of Nigeria showing the major regions of Nigeria and its
neighboring countries.

Nigeria is situated in the tropics with an average temperature of 900 F (32°C).
Annual rainfall varies from 98" (2.5m) in the southeast to 24" (0.6m) in the north.
Coastal forests cover the southern regions, giving way to savanna in the north. The Niger

‘River flows from the northwest to join the Benue River in central Nigeria, then turns
South forming a fertile delta as it empties into the Gulf of Guinea. About 35.6 percent of

tl'le land is arable and 12.4 percent is wooded. (Europa Publications, 1996; NewAfrica
Yearbook, 1997-98).

10

2.3 ECONOMY
The Nigerian economy is based primarily on agriculture and minerals.

Agriculture and Livestock accounted for about 35.3 percent of the Gross Domestic
Product (GDP) in 1992. Crude petroleum which accounts for about 12.9 percent of her
GDP and 97.9 percent of total export earnings makes Nigeria the sixth largest producer

of crude oil in the world and the second largest in Africa afier Libya. The development
of the petroleum industry in the 19608 and the 19703 radically transformed Nigeria from
an agriculturally based economy to a major oil-exporter.

The oil boom came at a time when Nigeria was concerned with the task of
postwar reconstruction. By 1966 oil was already Nigeria’s main source of foreign
exchange earning, accounting for a third of her total exports. Export earning from
petroleum generated high levels of economic growth, and by mid-19705 Nigeria’s

economy had become the dominant economy in sub-Saharan Africa. The oil boom had

fliree components, the increase in oil output in the early 19703, and the oil price shocks of

the 1973-1974 and 1979-1980.
The decline in agriculture was dramatic in the period from 1972 to 1985.

Production of rubber and cocoa, which had been growing at over 10 percent and 6 percent

Der annum respectively in the period from 1950 to 1966 declined to around 6 percent per
annum. By 1981 agriculture’s share of GDP, which was almost two-thirds in the periods

from 1960 to 1965, had fallen to 28 percent. Export crops declined in aggregate relative

‘0 services from 100 percent in 1966 to 11 percent in 1981.

ll

The decline in world oil prices from 1981 to 1986, led to a remarkable decline in
export revenue fiom crude oil. The oil shock was compounded by the unsustainability of
previous borrowing. Not only could the flow of borrowing not be sustained, but the
accumulated debt had to be serviced. As a result foreign exchange fell and with it a fall
in the imports of raw materials for her manufacturing industries. The fall in oil revenue
led to a drastic fall in economic activities with most of Nigeria’s industries Operating
below capacity as a result of shortages of imported materials and production and wages

plummeting. Cumulatively, the loses from the oil shock and the reduction in net capital
inflow during the five years were around 120 percent of Gross Domestic Expenditure

(Bevan, Collier and Gumming, 1992; Europa Publications, 1996; Oyewole, 1987).

2-4- POPULATION

By 1991 census figures, Nigeria’s population was estimated to be 88.5 million
811d by 1996 her population had grown to nearly 103.9 million, making her the most
Populous nation in Afiica and the eighth most populous nation in the world (Population
Reference Bureau, 1996). This represents about 15.4 million increase in population

over a five-year period, an annual growth rate of nearly 3.1 percent. This is among the

hi ghest even by sub-Saharan Afiican standards.

Her population is concentrated heavily in the southwestern, southeastern and the

t1(Dim-central parts of the country. Whilst still predominately rural, the population has
become more urbanized, increasing from 19 percent in 1963 to 24.9 percent in 1990,

making her the seventh most urbanized country in Sub-Saharan Afiica.

12

It is one of the most ethnically diverse countries in Afiica, about 395 mutually
unintelligible languages are spoken and roughly about 434 ethnic groups exist in Nigeria.
Many of these groups have only a few thousand members, with the result that Nigeria’s
population is dominated by the Hausa-Fulani, the Yoruba and the Igbo, concentrated in
the north, the southwest and southeast respectively. These three groups comprise about
57.8 percent of the total population (Bevan, Collier and Gumming, 1992; Myers, 1989)

In 1990-95, the estimated life expectancy at birth was 54.28 years for women and
52.50 years for men ( Keyfitz and Flieger, 1990). By world standards, this means that
mortality was somewhat responding to efforts by the international health community but
that mortality was still high and out of control. The Net Reproduction Rate in 1990-95
was estimated as 2.57 daughters per woman surviving to the mean age of childbearing
(Keyfitz and F lieger, 1990). This means that the next generation of daughters will be 157
percent larger than the current one, a very high rate of reproduction indeed. This study

will relate those demographic realities to the social and economic structure in which they

are embedded.

5. s. EDUCATION
Education in Nigeria is partly the responsibility of the state governments,

although the federal government has played an increasingly important role since 1970.

Primary education begins at 6 years of age and lasts for six years. Secondary education

begins at 12 years Of age and lasts for a further 6 years, comprising two six-years cycles.

Education to junior secondary level (from 6-15 years of age) is free and compulsory. In

13

1991 about 76 percent of school aged children attended primary school (79 percent of
boys and 62 percent of girls). In comparison, only about 23 percent attended secondary
school (24 percent boys and 17 percent girls). In 1993, about 54 percent of the adult
population were literate, about 64.7 percent male and 43.8 percent female. In 1994,
Nigeria had about 38 universities and education was allocated 4.1 percent of the total
expenditure in the 1990 federal budget. A once distinguished university system has

deteriorated due to repression and under funding (Europa Publications, 1996; NewAflica

Yearbook, 1 997-1998)

2-6. POPULATION POLICIES
For over three decades now, Nigeria’s population had been growing rapidly.

Policy makers have long recognized that Nigeria’s population was growing at a rapid
rate. However, the policy emphasis was economic grth and development first, and
men it was believed fertility decline could follow naturally. With the world’s Oil price
decline and the deterioration in Nigeria’s economy, it became apparent that something

drastic needed to be done about the rapid population growth. In 1988, an Official Family

P l arming Program was formulated and launched by the Federal Government through its

Ministry of Health. Nongovernmental agencies such as United Nations Fund for

Population Activities (UNFPA) and the International Planned Parenthood Federation
(IPPF) through its Nigerian affiliate, the Planned Parenthood Federation of Nigeria

(PPFN), operate family planning clinics in all states, supplementing the efforts of the

Federal Ministry of Health.

14

The population policy is predicated on the right of couples and individuals to
decide fully the number and spacing of their children, and the access to information.

education, and the means to exercise this right. The Objectives of the policy and its

implementation are described in details elsewhere ( Feyisetan and Ainsworth, 1994,

NDHS, 1992).

Current contraceptive prevalence rate that was only one percent among currently
married women a decade ago has since 1988 risen to 6 percent by 1990, 3.5 percent
modern and 2.5 percent traditional (NDHS, 1992). Since at the time of the survey in
l 990, official family planning had been in operation for less than two years, this is too
short a time for family planning policies to have a considerable effect on fertility

decisions of women of childbearing age. Furthermore, contraceptive use of 6 percent

among currently married women is still relatively low to result in a drop in fertility Of the

magnitude that data reveals.

15

9l

 

GULF OF GUINEA

 

ommtnmwmms.

 

I.” -~_

l‘A‘YE'I’ A R
P
U

a“ .
+1”
'5'.

. Hare. ‘0
I" .0

FIGUF

 

’ ’0 uman
t PLATEAU O m"

TARABA

    

1 CAMEROON

IAP OF NIGERIA

J ADAMAWA '

 

 

— Main railway
— Main road
— International boundary
—- Province boundary
4* International airport
0 Capital
0 Major town

(2) Other town

lanuary I998

 

0km 50
l l

100
r

150 200

 

l
0 miles

50

100\

 

 

CHAPTER 3

REVIEW OF LITERATURE

3.1 INTRODUCTION

 

This chapter has two objectives. First, it presents a review of the relevant
literature on the correlates Of fertility decline, particularly the relationship between
women’s education and. fertility. Second, it presents the hypotheses of the present study.

The inverse relationship between women’s education and fertility is well
established in the empirical research on fertility (Birdsall, 1988; Caldwell, 1980:
Cochrane, 1979; Schultz, 1997). Empirical studies have consistently point to education,
that of the wife and to a lesser extent that of the husband, as an important factor
accounting for fertility differentials within populations. Education, therefore has been an
important part in investigative work that concerns fertility differentials by socio-
economic status and also in the search for casual explanations of fertility levels and
changes in fertility. A knowledge of the link between education and fertility is especially
relevant for development planning because education can directly be influenced by
government policy (United Nations, 1987). There is some evidence from surveys
preceding both the World Fertility Survey (WFS) and the Demographic and Health
Surveys (DHS) of a connection between education and desired fertility. The WFS and
the DHS programs have greatly increased the amount of information available on the
relationship between education and fertility (United Nations, 1981; Ainsworth, Beegle

and Nyamete, 1996; Thomas and Maluccio, 1996).

17

Education, especially women's education can act both directly and indirectly
towards the reduction in fertility. By delaying the age at first marriage and at first birth,
schooling shortens the reproductive span of a woman thus limiting her potential supply of
children (Bulatao and Lee, 1983). Education may change attitudes, values and beliefs
toward a small family norm and towards a style of childrearing that is relatively costly to
parents in time and money (higher "child-quality that we discuss below). Education may
thus reduce marital fertility by altering a woman’s choices for the demand for children.

We will look at some Of the stylized facts about women’s education and fertility.

3.2 WOMEN’S EDUCATION AND FERTILITY.

First, schooling raises the opportunity cost of women’s time in raising children.
This is because education raises the marginal productivity of women in the formal labor
market, which raises the “price” Of children by raising the wage that women can earn in
the labor market. If time spent working at the labor market and time in child care are
mutually exclusive, then wage measures one of the principal opportunity costs of
childrearing. It is, therefore, likely that the higher the price of time, the lower should be
the desired-number of children (Becker, 1960; Willis, 1973). However, this may not be
as true in Africa, where work need not conflict with child care, since employment Of low
cost substitutes for the mother’s time, such as services provided by relatives, are readily

available ( Farooq, 1985; National Research Council, 1993).I

 

' Caldwell(l982) envisions a larger role for education than that considered in the simple economic model. In his view,
education serves as a vehicle for the adoption of Western ideas regarding the family. It encourages a more child-
centered view of one’s parental responsibilities. Education may alter the definition of what constitutes acceptable child
care, giving greater weight to the time spent by the mother with her child, compared to the time given by the mother
substitutes.

Second, women’s schooling effectively decreases the cost of obtaining
information concerning the use Of contraceptives. Education enhances the effectiveness
of obtaining information pertaining to the use of contraceptives. Thus the more educated
women can contracept better than the uneducated, reducing the number of unwanted
pregnancies. (Ainsworth, Beegle and Nyamete, 1996: Bongaarts, Frank and
Lesthaeghe, 1984; Thomas and Maluccio, 1996).

Third, women with more education have better access to information regarding
their own health and that of their children. The better educated woman is more likely to
produce healthier children, leading to a greater child survival ratio, as a result both infant
and child mortality declines. This makes it possible for couples to have fewer children to
achieve their ideal family size (Okojie, 1991; Preston, 1978; Schultz, 1997).

Fourth, there is the quantity-quality trade off argument. In any number of
societies that have experienced a fertility transition, changes in the perceived benefit and
cost of child schooling have played a key role in the transition. The impetus for fertility
change originates in the economic returns associated with schooling. In the course of
modernintion, an economy begins to display significant differentials in earnings by
schooling levels. Parents then come to regard schooling as an avenue to a better life for
their children and as a human capital investment that may, over the long term pay
dividends to the parents themselves. Yet education is costly in terms of both direct costs
and opportunity cost of foregone child labor. It generally remains too costly for parents
to give each child the desired schooling and continue to bear the customary number of

children. One element of household expenditures must give way, and typically fertility

l9

falls as investment per child increases (Becker and Lewis, 1973; Becker and Tomes,
1976; Montgomery, Kouame and Oliver, 1995; National Research Council, 1993; van de
Walle and Foster, 1990).

Finally, women’s education may also influence the within household distribution
of authority. More educated women may earn income for themselves and with it some
form of autonomy in the household. Consequently, they will become part of the
decision making process in the household. Decisions pertaining to fertility may thus be
made jointly, enabling women to make fertility decisions that may affect their own
health (Caldwell, 1980, 1986, 1995).

Some researchers, however, argue that female schooling may indirectly raise the
natural fertility level through its influence on improved nutrition, personal hygiene and
maternal health. Female schooling may also reduce the duration of breastfeeding and its
contraceptive benefits (Bongaarts, Frank and Lesthasghe, 1984; World Bank, 1984).
The breastfeeding, abstinence and mortality effects act to increase the supply of children,
raising the possibility that the net effect of education on the number of births or surviving
children may not always be negative. Education may also be associated with increased
.fecundability, lower pregnancy wastage and longer reproductive span for women through
better nutrition and health. Evidence that link fertility to nutrition and health is however,
sparse and conflicting. It is not clear that these latter factors have an important effect on

aggregate fertility (Bongaarts and Potter, 1983; Gray, 1993; Nag, 1983).

20

3.3 RESEARCH ON EDUCATION AND FERTILITY.

The relationship between female schooling and fertility may not be linear. The
inverse correlation between education is much more consistent for urban rather than rural
areas, the better educated rather than the least educated (Ainsworth, Beegle and Nyamete.
1996; Rodriguez and Cleland, 1980; Cochrane, 1979; United Nations 1987). Education
may bring fewer employment opportunities in the rural areas, the net cost Of children is
relatively lower in rural areas. On the other hand, high density, an industrial occupational
structure and greater exposure to many aspects of modernization in the urban areas,
implies high social and economic costs of children (Behrrnan and Deolalikar, 1988).

Studying 20 countries who participated in the World Fertility Survey (WFS),
Rodriguez and Cleland (1980) found that the total fertility Of rural residence exceeds that
of urban dwellers in all WFS countries. In Latin America and Caribbean, the differential
was as large for the married women as it was for all women, but in four countries the
disparity in marital fertility was noticeably smaller, indicating that later age at first
marriage among the urban population was a major differentiating factor between the two
sectors (Kenya, Jordan, Philippines and Thailand). In three countries the differential was
reversed, with rural women having lower marital fertility (Bangladesh, Indonesia and
Pakistan). There was also regional variation in the size of the differential, Latin America,
the Middle East, Kenya and a few countries in Asia (Korea, Philippines and Thailand)
have the largest differences of two to three children in the total fertility rate, and about

one to two children in the recent marital fertility. Other Asian countries and some

21

African countries for which data was available (Lesotho, North Sudan and Senegal) show
either much smaller differences or a reversal Of the marital fertility.

Using Demographic and Health Surveys (DHS) data from fourteen Afiican
countries, Ainsworth et al (1996) find differential impact of female schooling in urban
and rural areas. Fertility declines with increases in female schooling in both rural and
urban areas, particularly after primary schooling. The early years of primary schooling
are associated with higher fertility in urban Nigeria and Uganda, but are Otherwise
insignificant. The differential between women with eleven or more years of schooling
and those with no schooling is often greater in rural than urban areas (Cameroon, Kenya,
Niger, Nigeria, Uganda and Zambia). In general, at every level of schooling urban
women have lower fertility than rural women. At the highest levels of female schooling,
fertility declines more rapidly in rural areas and lower than urban areas in Nigeria.
Fertility in urban and rural areas converges however, at higher levels of female schooling
in Botswana, Kenya and Uganda. Differentials between urban and rural fertility when
controlling for female schooling are quite small in Botswana, Cameroon, Nigeria and
Zambia, but remains large in Ghana, Senegal and Togo.

It is sometimes argued that a negative impact of female schooling on fertility
emerges only when a certain minimum level of education is attained (Cochrane, 1979).
Rodriguez and Cleland (1980) find that in most Asian countries a few years of primary
education make almost no difference in marital fertility, and only secondary education is
associated with substantially lower fertility. However, in Latin America and the

Caribbean, any formal schooling, even a few years of primary schooling, usually result in

22

lower fertility, and both upper primary and secondary education are associated with
substantial reduction in fertility.

Using data on about 38 countries that participated in the WFS, fertility
differentials were also observed for women with differential educational status (United
Nations, 1987). Women with four to six or with seven or more years of schooling had
substantially lower fertility on average than women with less education. For half of the
countries the total fertility rate (TFR) for the highest educational groups falls in the range
of 2.6 to 3.4 children. In sub-Saharan Africa, education differentials in fertility tend to be
small and non-monotonic, with the highest fertility occuning in the intermediate
educational categories. In most cases, within each education group TFR exceeds the
fertility levels reported by older women. In Northern Afiica, Latin America and the
Caribbean and Asia and Oceania, the education differential in fertility is typically larger
and is more Often monotonic than in sub-Saharan Africa, and on average the differential
is widening. The mean TFR difference between the extreme educational categories is 3.1
children in Asia and Oceania and 3.6 in Latin America and the Caribbean. In Bangladesh
and Indonesia, fertility is higher in both intermediate and higher educational groups than
the uneducated women. In Latin America and the Caribbean, the education differentials
are usually monotonic. On average, the fertility difference between the two lowest
educational categories is 0.3 child for children ever born (CEB) and 0.6 child for TFR.
Women with seven or more years of schooling average 1.9 fewer children than those with

from four to six years according to TFR, and 1.3 fewer according to CEB. The fertility of

23

women with one to three years of schooling is frequently as high as, or slightly higher
than, the fertility of women with no schooling.

Ainsworth et a1 (1996) also using DHS data from fourteen African countries find
that the number of years of female schooling is significant and negatively related to
cumulative fertility in thirteen of the countries. They found that 1-3 years of schooling
is not related to cumulative fertility in twelve countries, and has a positive relationship
in two, Nigeria and Zambia. In half of the countries, women with 4 - 6 years Of
schooling have 0.2 to 0.4 fewer number of children than those with no education. The
gap widens to 0.8 to 1.8 fewer children ever born for those with eleven or more years of
schooling over those with no schooling. This suggests that only after certain thresholds
of education have been reached will fertility decrease. Using DHS data from 26 African
countries Martin (1995) also concludes that higher female education is consistently
associated with lower fertility. However, a considerable diversity exists in the magnitude
of the gap between upper and lower educational strata and the strength of association. In
some countries with low levels of female literacy, fertility tends to be higher among
women with small amounts of education.

In comparing fertility in Kinshasa (Zaire's capital) between 1955, 1975 and 1990,
Shapiro (1996) finds that female schooling at the secondary level and beyond ultimately
lead to a reduction in fertility. By holding age constant, fertility tends to be higher among
women who had received only primary schooling, followed by women with no schooling.
Among those who have received education beyond the primary level the average number

of children ever born generally declines steadily as educational attainment increases with

24

a sharpest decline among women with upper level secondary education compared to those
with lower levels of secondary schooling, and the lowest fertility among women with
university-level education. For example, their regression analysis reveals that relative to
women with primary schooling those with no schooling had on average 0.2 - 0.3 fewer
children born in 1975 and 1990. In comparison, those with lower secondary education
had about 0.5 fewer children. The fertility differential widens to 1.5 - 1.8 fewer children
for women with upper-level secondary schooling and to 2.1 - 2.5 fewer children for those
with higher education.

Shapiro and Tarnbashe (1994) using household survey data carried out in
Kinshasa 1990, they also find fertility differentials among women with different
educational attainment. Relative to women with primary schooling, women in all other
schooling categories, including those with no schooling, have consistently lower numbers
of children ever born. For instance, women with 1-4 years of secondary education have
0.52 fewer children. This compares with 1.4 fewer children for women with 5-6 years of
secondary education and 1.95 fewer children for women with university education after
controlling for age. When religion and employment status are controlled for, the
coefficients are 0.5 fewer children for women with 1-4 years of secondary education, 1.34
for women with 5-6 years of secondary education and 1.77 for women with university
education. They also found female schooling as a strong determinant Of other correlates
of fertility. For example, their empirical findings indicate that there is the tendency for
increased female schooling to be associated with a significantly greater likelihood of

using contraceptives. Schooling is highly significantly related to the use of modem

25

contraceptives, with university - level women showing by far the greatest likelihood Of
modern contraceptive usage. Women with upper-level secondary schooling are more
likely to have had an abortion than those with no schooling. Women who are enrolled in
school have a significantly lower probability of entering a first union than women with no
school enrollment.

Many reasons are given in the literature for the higher fertility of women with
some primary education over those with no schooling. Some argue that some primary
education, by exposing women to basic rules of hygiene and better dietary habits may
generate higher fecundity without influencing traditional family formation habits
(Arowolo, 1976). However, the research on the link between nutrition and fertility is not
conclusive. Bongaarts (1980) and Menken et a1 (1981) arguethat there is little empirical
support for the link between good nutrition and fertility. One may also argue that some
primary education by exposing women to literacy opens the door for better and well paid
husbands, and that the positive income effect from husband's education may outweigh
the negative effect of wife's schooling on fertility. Thus at the lower levels of schooling
the impact of schooling on fertility may be gender dependent. Ainsworth et a1 (1996)
argue that literacy may not be achieved in a year or two years of schooling and, under the
circumstances that prevail in most African settings, may not even be achieved until
completion of primary school. The small group of women who completed only a few
years of schooling are those who became pregnant, whose families wanted them to get

married, or simply could not keep up and therefore stopped their schooling.

26

Female education may also have a differential impact on fertility depending on the
woman's age. Some studies have found the strongest relationship among the middle-aged
women 25-34. At Older ages, especially for women in the age bracket 35-49, the
influence of education on fertility differentials diminishes greatly suggesting a late catch-
up effect on births at higher marriage durations ( Farooq , 1985; Thomas and Maluccio.
1996: Ware, 1975).

Okojie (1991) using data on Bendal state in Nigeria finds, however, an inverse
relationship between education and fertility even beyond the age 35. This also suggests
that education may have different effect on fertility depending on the age cohort Of the
woman, what is known as the "cohort effect" on fertility. The very young women 15-24
are too young and are just at the beginning of their childbearing years and some may not
have completed their education, so the influence of schooling on fertility may be
insignificant. On the other hand the Oldest women 35-49 have completed their education
and are very close to the end of their childbearing years. The incentive to delay births for
more education may not be great. In addition, for biological reasons their capacity of
having more births is greatly reduced, so again extra schooling may not alter their
preferences for children that much. The greatest impact of female schooling is on the
middle cohort, 25-34, the most fecund group. These are in their prime childbearing
period and most of whom are still engaged in schooling or at the end Of their schooling.
Education is beginning to pay Off occupationally so the Opportunity cost of childbearing

may be quite high. Thus the marginal effect of education on fertility is most significant

27

for this age group. Thus the “cohort effect” may interact with female schooling
producing the differential effect of education on fertility.

Female education also has a differential impact on the timing of first marriages
and first births. Using Zimbabwean Demographic and Health Survey, 1988 data and the
Situation Analysis Study, 1992, Thomas and Maluccio (1996) found that women who
complete primary school have their first child about seven months later than those with
no schooling. Women who complete eleven or more years of schooling wait another two
years before having a child. They found the biggest effect among women with twelve or
more years of schooling, they delay childbirth for five years relative to those without any
schooling. They also found that birth spacing tends to rise with education and the effect
‘ is significant only among the better educated younger and urban women.

Cleland and Chidambararn (1981) also using WFS data found differentials in the
female education-fertility relation in terms of the tempo of childbearing at different
marital durations. Fertility is uniform across educational groups in the early stages of
marriage but begins to diverge from the duration 5-9 years onward as better educated
groups use contraceptives to restrict family size. This pattern was representative of Peru
and other Latin American countries, and several other Asian countries as well. However,
for Indonesia, Kenya and Pakistan, there is a positive education-fertility relationship at
early marital duration, reverting to a negative relationship at durations greater than 10
years.

The positive association between female schooling and contraceptive use is also

well established in the empirical literature. Ainsworth et a1 (1996) found that urban

28

women and women with more schooling have higher rate of contraceptive use than the
rural and less educated women. The major increase in current contraceptive use occurred
with early years of female schooling in Burundi, Ghana, Niger, Senegal and Tanzania.
However, marginal increases in the use of contraceptives for additional female schooling
is relatively smaller even for countries like Botswana, Cameroon, Kenya and Zambia,
which record high relationships between contraceptive use and female schooling.

Apart from education, there are also social and community variables that may
affect a couple's desire for children. Notable among these are race and ethnicity, religion .
and type of union. Research indicate that women in polygamous marriage on average
have fewer children. Namfua (1981) in studying polygyny in Tanzania, found the
average mean age of a woman in a polygamous marriage to be higher than that of women
in monogamous marriage. In addition the age at first marriage for women in a
polygamous marriage was also higher than their counterparts in monogamous marriage.
On average women in polygamous marriages have longer birth intervals than their
counterparts in monogamous marriage. When schooling is controlled for, the results
indicates that for women with no schooling, about 82.6 percent are in polygamous
marriage, whereas only 69.6 percent are in monogamous marriage. In comparison, for
women with 5-8 years of schooling only 4.4 percent are in polygamous marriage and 10
percent are in monogamous marriage. None of the women with twelve or more years of
schooling is in polygynous union.

A couple’s religious status connotes a system of values that can influence fertility

via two routes: 1) directly, by imposing sanctions on the practice of birth control or

29

legitimizing the practice of less effective natural methods only, or 2) indirectly, by
indoctrinating its followers with a moral and social philosophy of marriage and family
that emphasis the virtues of reproduction (Westoff, 1959). Substantial religious
differentials in fertility have been empirically documented in large number of countries.
For example, in Lebanon, Egypt, Israel, Soviet Union, Jordan, India and tropical Afiica
(Yankey, 1961; Rizk, 1963, 1973; Matras, 1973; Mazur, 1967; Sinha, 1957; Caldwell,
1968). All these studies have found significantly higher fertility rates for Muslims than
for non-Muslims. In the West, religious affiliation has been found to have a significant
effect on fertility. For example, religious fertility differential have been found in Europe,
Canada, United States, South Afiica, Australia and New Zealand. Studies have shown
that Catholics have higher fertility than non-Catholics (Glass, 1968; Chou and Brown,
1968; Nixon, 1963; Burch, 1966; Ryder and Westoff, 1971; Higgens, 1964; Day 1964,
1968). Usually Catholics have been found to have higher fertility than Jews or
Protestants, with Jews having the lowest fertility among the three groups.

However, in some rural areas the religion-fertility relationship is somehow mixed.
For instance, Yankey (1961) noted similar fertility levels for Muslims and Christians in
rural areas of Lebanon. Rizk (1963) also found this to be the case in rural Egypt. Busia
(1954) noted no difference between Muslim and Christian fertility in Ghana.

Women’s education may also have a different impact on fertility depending on
one’s racial and ethnic background. In countries such as South Africa where the
population is divided among sharp racial and ethnic lines, the relationship between

education and fertility differs accordingly. For example, using multivariate regressional

30

analysis. Thomas (1996). finds the greatest impact of female education on fertility
among black and colored women. Among the black women, the negative association
between fertility and education is relatively weak at the bottom of the education
distribution and is significant only among the women with at least three years Of
schooling. Among the colored, fertility appears to rise with women’s education until
around four years of schooling (standard 4). However, among colored women with more
than five years of schooling there is a powerful negative association between education
and fertility. For the Indian women, fertility increases with education until around the
completion of primary school and then falls dramatically until ten years of schooling at
which point fertility tends to stabilize with female education. They made similar
observations for the white women as they did with the Indian women. For example in his
study, he finds that a year of female schooling reduces fertility by 0.12 among black
women in South Africa. The correlation is almost as large for the colored, about half as
large for Indians, and only about 0.04 for white women.

Abu Bakar et a1 (1985) using a Household Survey data, 1979, also found fertility
differentials among the three major ethnic groups in Malaysia. After holding constant
ethnic and demographic variables, Malays and Chinese consistently have higher
cumulative fertility than Indians. For example, among all the female sample, the recent
fertility recorded by Malay women is 1.16, that of the Chinese is 1.05 and for the Indian

is 0.96.

3l

3.4 RESEARCH STUDIES ON NIGERIA

 

Interest in African fertility issues has stirred up a tremendous amount of research,
especially with the availability of data from countries participating in the DHS data
collection. Nigeria is one such country that has attracted research on fertility issues, in
part because there had not been any reliable census data since the civil war from 1967 to
1970. However, with the availability of data from DHS, the adoption of a National
Family Planning Program and compulsory primary education, Nigeria has become the
focus of recent research.

Using NDHS , 1990, Feyisetan and Ainsworth (1994) examined the
determinants of contraceptive use and fertility in Nigeria. They found that female
education and contraceptive use are positively related and that the more educated women
have fewer children. The influence of education on contraception is higher for the rural
areas than the urban areas, and in the Northern areas compared to other areas. They also
found a strong negative association between female schooling and fertility. An
additional year of female schooling is associated with the greatest increase in
contraceptive use in rural Southwest and rural Northwest and smallest increase in the
urban Southeast and Northeast. The marginal effect of an increase in female schooling of
one year on current contraceptive use is 0.13 in urban Southwest and 0.24 in rural
Southwest. It is 0.14 in urban Northwest and 0.24 in rural Northwest. Similarly it is 0.03
in urban Southeast and 0.14 in rural Southeast, 0.05 in urban Northeast and 0.15 in rural

Northeast.

32

In estimating the fertility response to child survival rate in Nigeria, using Bendel
State as a case study, Okojie (1991) finds an inverse association between female
education and children ever born for women who are over 35 years Of age in her
multivariate analysis. Female fertility declines monotonically with higher levels of
education, even after controlling for age-education interaction. The negative relationship
between education and fertility is highly significant among women in the age group 35-
50 but insignificant for the younger age group 15-24. She finds child survival ratio
negatively related to fertility for all women and for all age groups. It is more significant
for women in the age group 25-34, women who are in their prime childbearing period.

Adewuyi and F eyisetan (1988) using a survey of mothers residing in Lagos,
analyzed fertility differentials among three major ethnic groups, the Igbos, the Yorubas
and the Hausas residing in Lagos. They found in their bivariate analysis that for all
ethnic groups, women with secondary education have the lowest mean number Of
children ever born upon arrival in Lagos. For example, among the Hausa, women with
no education have on the average 5.1 mean number of children ever born (CEB), those
with primary'education have 3.3 and those with secondary and above have 2.7. Similar
results are found among the Igbo and the Yoruba. Schooling made a difference in
fertility both in the pre-arrival and post-arrival number of births. They also found that
urban life has a depressing effect on fertility. Respondents acknowledged that raising
children in the urban setting was more expensive. They also found fertility differentials

among the different ethnic groups. For instance, among the three ethnic groups residing

33

in Lagos, the Hausas had 3.9 mean CEB, the Igbos had 3.6 and the Yorubas 3.8 mean
CEB.

Ebigbola and Omideyi (1988) also analyzed the fertility behavior in an urban city
using ever- married women in Ilesa, Oyo State, Nigeria. They also found a negative
relationship between female education and the mean number of CEB. For example, on
average, women with no education have 5.5 CEB, women with primary education, 4.0.
those with secondary education 3.5, and those with post-secondary education 2.7. This
Observation was made for all age groups except for women in the age group 45-49 where
those with secondary education have higher number of births than those with primary
education.

Using a survey Of women in Ibadan, Arowolo (1976), analyzed the determinants
Of fertility amOng ever- married Yoruba women aged 15-49 in Nigeria. His bivariate
analysis suggests that female education has a depressing effect on fertility. For example
women with no schooling had on the average 4.1 mean number of CEB, those with
primary had 3.7, those with high school education have 3.5 and those with University
education had 3.1 mean number of CEB. This Observation was made among all women
in the different age groups with the exception of those in the 40-44 age groups where
women with no schooling have lower number of births than those with any level of
education.

Orubuloye (1981) using a survey of nrral women in Ibadan and Ekiti Division of
Western Nigeria, found a negative relation between fertility and level of female

schooling. For example, in Ibadan, the results indicate that in almost all age groups

34

(except women ages 30-39), women with primary education had lower cumulative
marital fertility than women with no schooling. However, in Ekiti villages, it was only
women ages 30-44 with primary education who had fewer cumulative marital fertility.

Orubuloye (1981) also found that in both villages and for all women age groups,
women in polygamous maniage had fewer cumulative marital fertility than women in
monogamous marriage.

Obadike (1968), found that currently married Muslim and traditional worshipers
in Lagos were 10.5 percent more fertile than the Christians, and that Catholics were
slightly less fertile than Protestant and far less so than Muslims. In contrast Lucas
(1976) investigation among Lagos women show that Yoruba Christians reported a higher
mean number of births than Muslim counterparts. Similarly, Christians in Ibadan were
also found to have a higher mean number of births than the Muslims, this was particularly
so among the older women (Sembajve, 1981). It has also been found that Christians in
Ekiti villages have lower fertility than non-Christians. In contrast, the non-Christians in
the Ibadan villages appeared to have smaller number of children than the Christian
population (Orubuloye, 1981). It is also argued that Muslims amongst the Yorubas are
very different from the Northern Hausa Muslims in fertility matters.

It has been widely argued that the social environment in which the people lives is
a paramount factor in fertility analysis (Freedman, 1975). Fertility is a complex
phenomenon and may vary greatly from one population to another regardless of the
socio-cultural identification of individuals. Birthplace is also known to influence a

woman's fertility. A survey of Yoruba immigrants women shows that they have a lower

35

mean number of births than the Yoruba who were born in Lagos or have lived in Lagos
for 10 years or more (Lucas, 1976)

Nearly universal marriage and long period Of postpartum infecundability appear to
be the major determinants of fertility among the Yoruba. In Orubuloye (1981) analysis,
approximately 80 percent of the total fertility variation is explained by the effect of non-
susceptible period, while the effect of non-marriage accounts for the remaining 20 percent

of the variation. Contraception and induced abortion had no measurable impact.

3.5. PROBLEMS ASSOCIATED WITH EMPIRICAL WORK ON FERTILITY.
The research work on female education and fertility is extensive. Most
researchers use the total number Of CEB as the dependent variable with a number of
regressors including female education, age and age squared, husband’s education along
with other community variables which affect fertility exogenously. There is a problem
with this approach that the present paper seeks to address. As mentioned by Ainsworth,
Beegle and Nyamete (1996), different women in the survey have different durations of
exposure to childbearing. The younger women have not completed their fertility, so this
creates a censoring problem. One way of dealing with the problem as they suggest is to
predict the completed fertility of the younger women based on the completed fertility of
the older women (Ainsworth et a1, 1996). The problem with this approach is that it will
restrict the analysis to the older women who are the least educated and confound life-

cycle with historical changes in the relationship between fertility and education. The

36

comparison of fertility over different age cohorts is lost when using the mean number of
children ever born.

The approach we propose in this study is to take a snapshot Of the women, in
particular birth cohorts at different ages over the years and look at how fertility is
changing with education across and within cohorts. For example, for each woman in the
sample we calculate how many children had been born at ages 18, 23, 28, 33, 38 and
43. For each of these variables we look at the relationship between fertility and education
and also how fertility is changing from one cohort to another overall and within
education groups. We reconstructed the cumulative fertility at various attained ages
using the retrospective birth histories of women in the survey. Comparison of fertility
across cohorts, which can not be done when using total number of children ever born by
survey data, can thus be done with this approach, although we have assumed no time
effect, since cohort, time and age effects are not all simultaneously identified. The added
advantage is that for each level of cumulative fertility, the women in the relevant sample
have equal exposure to the risk of childbearing. The age at which the relationship
between education and fertility diminishes can also be captured easily.

Research in this line has been undertaken by Lam, Sedlacek and Duryea (1992)
using Brazilian data and there exist numerous time dynamic studies as well. From the
birth histories, they reconstructed cumulative fertility at various attained ages. Using
both descriptive and multivariate analysis, Lam et a1 (1992) were able to show that
female education and that of their spouses accounted for about 40-80 percent of the

fertility decline in Brazil. Our descriptive and regression analysis will fully capture the

37

predictive influence of female education on fertility and also to help find out if fertility
has fallen and over which level of education. We will also explore for what cohort the
influence of female education on fertility diminishes. The multivariate analysis fully
captures the dynamic relationship between female schooling and fertility controlling for
age cohort, husband's education and community variables such as area of residence and
religious affiliation and type Of marriage Of the respondents.

Furthermore many Of the research on Nigeria cited above restrict their analysis to
a particular region of the country, which may not be representative of the broad spectrum
of the Nigerian population. Some of them also restrict the study to ever-married women.
These women are at a higher risk of pregnancy than the unmarried, however childbearing
outside marriage is an acceptable practice in African societies including Nigeria, so
looking at fertility for the unmarried women is also Of concern for policy issues. The
NDHS, 1990 covers the whole of Nigeria and looking at the fertility issues for the whole
country and the added advantage of comparison of fertility among the different areas and
regions of residence is something worth researching. This is the gap we hope to fill in the

empirical literature using Nigeria as a case study from Africa.

38

CHAPTER 4

POTENTIAL MEASUREMENT ERRORS IN THE NIGERIA DEMOGRAPHIC
AND HEALTH SURVEY, 1990.

4.1. INTRODUCTION.

 

Survey design and implementation techniques have improved greatly over recent
years. Greater sophistication and accurate demographic measures have also evolved over
the last several decades. The Demographic and Health Surveys have incorporated the
latest innovation in survey design and research techniques andthe state of the art
demographic measurements. Nevertheless, all survey data contain measurement errors
that can affect demographic estimates. Before any conclusions can be drawn from any
survey one needs to assess the validity and reliability of the survey data (Andoh, 1980;
Arnold, 1990).

The Demographic and Health Surveys are a primary source of data for both
fertility and mortality analysis. Reliable fertility estimates heavily depend on complete
and accurate reporting of age of respondents and the total number of live births they have
had. Accurate reporting Of age of respondents, age of the children ever born, age at first
marriage and age at first birth are particularly important, since many of the estimates
derived from surveys, especially fertility rates depend heavily on age. Fertility estimates
may be affected by underreporting or overreporting of number of children ever born as
well as errors regarding date of birth or age (Arnold, 1990).

Unfortunately, age misreporting is a major source of measurement error in most

censuses and surveys from Afiican countries. In populations where reliable vital

39

statistics are not available and where ages and dates are not required daily, individuals
Often lack documents indicating age. Most problems of age misreporting are due to the
respondents’ ignorance of their exact ages and ignorance Of the ages of their children.
These problems are even compounded further in the African situation by low levels of
education among the oldest women (Ewbank, 1981; Rutstein and Bicego, 1990)

The literature on the origins of age misreporting also suggests that unknown ages
of respondents are usually estimated by interviewers on the basis of such personal
characteristics as the appearance of the respondent, marital status, the number of children
the respondent has ever born alive, the age of her last born child and so on. However,
these personal characteristics are very subjective.

This chapter looks at some of the potential measurement errors in the Nigerian
Demographic and Health Survey (NDHS), 1990. In what follows we look at the
accuracy Of the reporting of date of birth Of the respondents and their birth histories.
Section 11 looks at the data and how information on age and birth histories was collected.
Section III first looks at some of the problems reported by the staff on the NDHS.
Second it discusses some of the measurement errors we encountered when using the data
in our descriptive chapter on fertility trends in Nigeria. Finally in Section IV we

conclude the chapter.

40

4.2. DATA COLLECTION

 

The data for the present analysis are derived from the 1990 NDHS. This is a
national random sample of women ages 15-49. The NDHS was a joint project
undertaken between the Federal Government of Nigeria, which is represented by the
Federal Office of Statistics (F OS) and the USAID. The survey was conducted by the
PCS with technical support from the Institute for Resource Development (1RD) Macro
International Inc. Overall 5,999 households were interviewed and about 8,781 women of
reproductive age were interviewed and information on the health of their 8,1 13 children
under five years of age was collected. The survey was designed to elicit information
from the respondents on birth histories, marriage patterns, the use of contraceptives and a
few socio-economic characteristics, among other things. Three questionnaires were
administered in the main fieldwork for the NDHS: the household questionnaire, the

individual questionnaire, and the service availability questionnaire.‘

Age Reporting

For every woman in the survey information was sought on her age. Respondents’
ages in the individual questionnaire were obtained by asking for both the date of birth
( year and month) and age in completed years. The interviewer was instructed to ‘make a
serious effort’ to reconcile the answers if inconsistent. If the respondent could not give
either her date of birth or age, the interviewer was instructed to request a document

indicating the respondent’s age or to probe by estimating the respondent's age in relation

 

1For a complete description of sampling procedures see NDHS final report 1992.

41

to the ages of other members Of the household. Or the interviewer was asked to estimate
the respondent’s age in relation to the date of her first maniage or first birth. If all else
failed, the interviewer was instructed to guess the respondent’s age or age was estimated
by imputation.2

In societies where the level of education of the respondents is very low and
displacement of documents indicating date of birth is common, DHS interviewers in
addition to detailed questioning and probing techniques have resorted to historical
calendars. Historical calendars are individualized calendars of major historical events
that have happened in the history of the country, such as major earthquakes, famine, civil
war, and so on. This is to help respondents to remember their date of birth and help
improve the memory Of the respondents. However, there was no use of historical
calendars in the data collection process used by the field staff of the NDHS, 1990

(personal communication with Bert Themme).3

 

2 For key events in the respondent's life, dates have been imputed when full date of the event was not
provided by the respondents or in some cases if dates are inconsistent. These events are the date of birth of
the respondent, the date of birth of each child of the respondent, the date of conception Of current
pregnancy, and so on. All dates in the imputation process are used in the form of a Century Month Code
(CMC) which is the number of months from the start of the century to the month of the event. lmputation
is done in three stages. The first stage an unconstrained range is constructed from the available
information. If month and year are both available, the upper and lower bounds of the range will be the
same. If only a year is available, the unconstrained range will cover 12 months. If no years is given the
unconstrained range will cover the full range of possible dates ( 50 years before the survey). In the second
stage, ranges are adjusted to satisfy certain isolated constraints (constraints affecting the date in question).
In the third stage, ranges are adjusted to satisfy neighboring constraints (other events in the respondent's
life). At the end of the process the logical range should be either negative, zero or positive. If negative,
the date is inconsistent, and is corrected manually. If it is zero the date is consistent, and if it is positive
the date is consistent but not complete. For a complete description on imputation procedures, see the DHS
processing manual (Institute for Resource Development, 1989).

3Bert Themme works at the DHS head Office.

42

Birth Histories

The respondents were asked to report all live births, including births of children
who have died, first in terms of the aggregate number of children ever born (CEB), and in
terms of specific questions about each live birth (birth history). The CEB data were
collected in a series of six questions about the number of sons and daughters who are
living at home, the number Of sons and daughters living elsewhere, and the number of
each sex who has died.

The birth history data were collected in chronological order, starting with the first
birth and ending with the most recent birth. Data were collected on the date of birth, sex,
survivorship status, current age and whether living with the mother (for living children)
or age at death (for dead children). Following the birth history, the interviewer checked
the consistency of the CEB and the birth history and reconciled the differences. When
complete information on year and month of birth was not reported, the dates were
assigned by imputation. Once again the use of historical calendars would help
respondents, especially the older women and the least educated to put the dates of births
of their children in historical perspective, but this procedure was also not used in the

collection of the birth histories.

Age at First Marriage
From a sociological point of view, marriage may be defined as a “legally
sanctioned relationship involving economic cooperation as well as normative sexual

activity and childbearing, that peOple expect to be enduring,” (Macionis, 1997).

43

However, the Afiican conception of " customary marriage" is a “union of a man and a
woman, for the duration of the woman’s life, being normally the gist of a wider
association between two families or set of families.” (Atado,1988; Orji, 1983).

Under the Nigerian legal system, four kinds of marriages exist: the Customary
Law, Islamic Law, Christian Law, and Statutory Law. The Customary and Islamic Laws
promote polygamy, whereas that of Christian and Statutory Laws promote monogamy.
The primary purpose of marriage in the African context is for procreation and secondarily
for companionship( Atado,1988; Orji, 1983). The emphasis on procreation is so great
because of cultural and religious reasons such that childbearing outside of marriage is an
acceptable act in society. This situation sometimes confounds data collection that probes
into when marriage legally started. Is it the time when the customary rights began or
when the two partners started living together? In the NDHS, 1990, taking cognizance of
the existence of informal unions, marriage was defined as legal or formal marriage or
cohabitation. Legal marriage was recognized when couples were living together under
customary marriage or maniage under any of the other three coexisting marital laws, and
cohabitation was recognized when the respondent started living with her partner even
though no formal marital ceremony had been performed.

Information on marriage was collected in a series of questions. Women were first
asked whether they have ever been married or have lived with a man. Those who report
having had a husband or partner were then asked about their current union status and
whether they had been in one or more than one union. Finally, respondents were asked

to provide the month and year they started living with their first husband or partner. 1f

44

the woman could not provide a year of first union, she was then asked to give her age at
the time of the first union. Therefore, errors in measuring a woman's age at interview are

correlated to some extent with those in measuring her age at first union.

Age at First Birth

Age at first birth is seen as a more appropriate indicator of the beginning of
exposure to childbearing than age at first marriage, especially in countries where informal
unions and premarital births are acceptable. lnforrnation on a respondent’s childbearing
history was gathered by two procedures in the DHS surveys. First, each respondent was
asked about the number of sons and daughters she had living with her, sons and daughters
living elsewhere, and sons and daughters who had died. Then she was asked to provide a
full liVe birth history in which the name, date of birth, sex, survival status, and age at last
birthday, or age at death, was collected for each birth

Age at first birth is calculated by subtracting the woman’s birth date from the
birth date of her first child. Thus the accuracy of information on age at first birth depends
on the accuracy of the respondents reporting of both her own date of birth and that of her
first child. When the respondent’s own date of birth and that of her first born are

missing the information was imputed.

4S

4.3. MEASUREMENT ERRORS
In this section of the paper we present some of the measurement errors we found
in the NDHS, 1990 concentrating primarily on age of respondents and their birth

histories, age at first marriage and at first birth.

4.3.1 Age Heaping.

 

Respondents Age

Age heaping is the tendency to report ages ending in certain preferred digits
especially, zero or five. The greater the amount of this ‘age heaping’, the lower the
confidence in the quality of age data. Fertility analysis is very age-specific, and age
heaping may lead to serious biases in the estimation of age-specific fertility rates or age-
specific mean number Of children ever born. The problem of age-heaping is thus very
crucial in our fertility analysis in chapters 4, 5 and 6 of this present study.

According to Figure 4.1, there are age spikes at 15, 20, 25, 30, 35, 40 and 45
years of age. For example about 7.2 percent of the women said that they were 20 years
Old and as high as 9.5 percent said they were 30 years of age. In all as Table 4.1
demonstrates, about 21.4 percent of all the women reported their ages with figures ending
in five (15, 25, 35, and 45), in addition about 22.6 percent also reported their ages ending
with zero (20, 30 and 40).

To understand the extent of this age heaping we decided to calculate the
Whipple's Index for age concentration on the digits “0" and “5" for the age group 18-49 .

The Whipple’s Index is a measure of the reliability of the recorded age distribution It is

46

calculated by summing the population between the ages 23 and 62 whose ages end in “0"
and “5" and dividing the result by one fifth of the total population between 23 and 62, and
then multiplying the result by one hundred. The result should vary from 100,
representing no concentration on “0's” and “5's” to a maximum of 500, representing
absolute concentration of the entire population at these ages. Between these two
extremes the following categories for estimating the reliability of age distribution have
been recommended by the United Nations : a) Highly accurate - less than 105 b) Fairly
accurate -- 105 - 109.9 c) Approximate -- 110-124.9 d) Rough 125-174.9 and e) Very
Rough --1 75+.4 The Whipple’s Index for the NDHS, 1990 was 224.6 which indicates
that age reporting by the respondents was very inaccurate.S

When dealing with fertility analysis the traditional age grouping has been 15-19,
20-24, 25-29 and so on. If evidence, however, suggests age heaping then centering age
groups on preferred digits would prevent biases in demographic estimation based on age
groups. Since the NDHS has serious age heaping problems as indicated by the Whipple's
Index, we decided to regroup the age variable centering on “0" and “5" in our fertility
analysis. To give equal probability Of occurrence for both digits we used the women in
the age group 18-47, so for our five-year age groupings we had 18-22, 23-27, 28-32 and

so 011.

 

4 See United Nations Demographic Yearbook, 1992. Since our data covers women ages 15-49, and to
make sure both digits “0" and “5" have the same chance of inclusion we calculated our Whipple's Index for
women in the age group 18-49.

5 The Whipple's Index is after imputation, since reported ages as is in the data is after all imputation has
been done.

47

Age at First Marriage and Age at First Birth

According to Table 4.1, there is also evidence Of heaping of age at first maniage
and also at first birth. For example, about 2.9 percent of the respondents reported to have
been married at age 10, as high as 15.4 percent gave their age at first marriage as 15, and
7.1 percent at age 20. We do not see much of heaping at 25 or age 30.

Age heaping at first birth is not that pronounced. We see some heaping at age
15, about 10.1 percent of the women in the survey reported to have had their first births at
15, but similarly high percentage Of 10.6 reported to have had their first births at age 17.
Age heaping in the age at first marriage and at first birth is normally hard to detect
because first marriage and at first birth are usually highly concentrated in a narrow age
range. TO see the extent of age heaping at first marriage and first birth we decided to
calculate the Whipple's index for the reporting of age at first maniage and first birth. The
Whipple's Index for age at first marriage was 121.73 and that for age first birth was
— 112.78, which by the United Nations standard is approximate. This indicates that age
heaping is not as much of a problem in the age at first marriage and at first birth as with
the respondents' age.

There was another problem with the reporting of age at first marriage and first
birth. First, about one percent of the respondents reported to have been married before
age 10, and about 3.8 percent reported to have had their first births by age 12. This could
be real or pure measurement error especially in the African setting where young girls
could be betrothed at even an infant age but do not coreside until much later. For

example, in most arranged marriages, it is possible for girls to be betrothed at an infant

48

age. In such circumstances it is difficult to know when actual marriage started, that is
whether marriage is defined when betrothal took place or when the couple started living

together. However, this argument may not hold for age at first birth since this cannot be
misinterpreted so the problem with age at first birth at age ten and less could be real or

pure measurement error.

49

83 .3322 .fleoeeoemom me 8335me one. -3. 853m

 

 

— COS-O; no comm—
ou<
8339:33333&5323223
.o
.F
.u
.n
.c
:c
to
.. a
.4
Ho

 

0..

name)“ 30 red

50

TABLE 4.1- Distribution of Age of the Respondents, Age at
First marriage and Age at First Birth.

 

 

Age of % Age at First % Age at First “/c
Respondent marriage Birth
15 4.4 6 0.1 6 0.0
16 3.7 7 0.2 7 0.0
17 3.8 8 0.3 8 0.0
18 4.1 9 0.4 9 0.1
19 3.1 10 2.9 10 0.7
20 7.2 1 l 2.2 l l 1.0
21 2.6 12 6.7 12 2.0
22 3.8 13 7.2 13 3.3
23 2.8 14 1 1.3 14 5.2
24 2.7 15 15.4 15 10.1
25 8.0 16 8.0 16 9.2
26 2.8 17 7.6 17 10.6
27 2.8 18 7.8 18 9.3
28 3.2 19 6.6 19 9.2
29 2.1 20 7.1 20 9.7
30 9.5 21 4.2 21 6.2
31 1.3 22 3.4 22 5.8
32 2.0 23 2.3 23 4.1
33 1.4 24 1.7 24 2.9
34 1.6 25 1.8 25 2.5
35 5.4 26 0.9 26 2.1
36 1.3 27 0.7 27 1.5
37 1.3 28 0.5 28 1.0
38 1.8 29 0.2 29 0.9
39 1.2 30 0.3 30 0.6
40 5.9 31 0.1 31 0.4
41 0.7 32 0.0 32 0.4
42 1.3 33 0.0 33 0.2
43 0.7 34 0.0 34 0.3
44 0.8 35 0.0 35 0.1
45 3.6 36 0.0 36 0.1
46 0.7 41 0.0 37 0.1
47 0.7 42 0.0 38 0.0
48 1.1 39 0.0
49 0.7 40 0.0
41 0.0
45 0.0
46 0.0
8781 100.0 7080 100.0 6477 100.0

 

51

4.3.2 Incentives for Misreportingfl Enumerators

Recent claims of a fertility transition in certain countries in Africa are met with
skepticism by some researchers because data collection in tropical Africa gives rise tO an
appearance of fertility decline in the years immediately prior to a survey (Caldwell and
Caldwell, 1992). Arnold (1990) reports that interviewers in their attempt to decrease
their workload may be inclined to change the birth dates of selected children so that
children may not be included in the survey. For example, with the Nigeria Demographic
and Health survey, 1990, immunization and other health related questions were asked
about children who were born since January 1985, that is children born within five years
of the survey. If a child is born six or more years before the survey, then the over 48
questions on health related issues would not be filled by the interviewer. Thus
interviewers who want to cut down their workload are more likely to displace births
which occurred within the fifth year before the interview. To analyze the interviewer
effect on birth displacement, we turn to the NDHS, 1990 final report on the reporting of
birth and data quality.

Analysis is this section is based on Table D4 and Figure 3.2 Of the NDHS, 1990
final report (see NDHS, 1992, pages 28 and 163 ). In Figure 3.2, the dots represent
births in single calendar years and the solid line is a profile of a five-year moving
average of the number of births. The profile rises uniformly and then levels Off sharply
in the five years preceding the survey instead of continuing to rise. The single births
which are represented by the dots are very close to the profile for moving average except

for the years immediately closer to the interview. We see the greatest deviation of single

52

births from the moving average profile in the fifth and sixth years before the interview.
This evidence is also supported by Table D4, notice in particular that the total number Of
children born (both living and dead) was about 1,730 a year before the survey. It
fluctuates in the 1,6005 the second to the fourth years before the survey. It then drops
sharply and reaches 1,434 the fifth year before the survey, and rises abruptly to its peak of
2,072 the sixth year before the survey. There is therefore, evidence to support the fact
that there is underreporting of births the fifth year before the survey and overreporting Of
births the sixth year before the survey. Interviewers are likely to displace births in the

fifth year to the sixth year before the survey.

4.3.3. Underreportifl due to Memory Lapse by Respondents

 

As with any data where retrospective information is required, the accuracy of
answers the respondents give is very important. Respondents may deliberately provide
wrong answers for cultural or religious reasons. Answers to questions may be wrong
because the respondent’s memory fails, or probably she may not know the right answer.
Response errors arising from memory lapses affect greatly questions which have either
time reference or questions which refer to events that took place in the distant past.

Another source of response errors may be from the forward age-telescoping,
which is known in the literature as forward displacement of events which occurred in the
distant past. Respondents tend to report their date of birth and ages closer to the survey
period than they actually occurred. If this is the case Older women may consistently

underreport their ages and that Of their children. To investigate the extent to which

53

response errors by the respondents were affected by memory failure or age telescoping,
we decided to look at the completeness of information making reference to age of the

respondents and the accuracy of their birth histories

Completeness of Information on Date of Birth of Respondents

In Table 4.2 we begin to document the completeness of information on the date of
births given by the respondents in the NDHS, 1990. According to Table 4.2, for all the
women in the sample, on average only about 60 percent provided complete information
on their date of birth. The percentage of women with a complete date Of birth is about
75.4 percent for the youngest cohort 15-17. This falls steadily from the youngest to
about 44.3 percent for women ages 38-42, and then rises for the next two oldest cohorts.

Female schooling, however, improves the memory of the respondents. For
example, in the lower panel of Table 4.2 we look at the completeness of age reporting by
years of schooling. Among women with no schooling, on average about 42.4 percent
provided complete information on their date of birth. This falls, sharply from 53.9
percent for women ages 15-17 to 44.9 percent for women ages 18-22, then continues to
drop gradually to 35.4 percent for women in the 38-42 age group, then rises with the
older women 43years and over. Cell size is not a problem in this schooling group so we
can not use that as the argument for sudden improvement in memory for women ages 43-
49, this may be a sign of misreporting by the oldest cohort, since memory lapse possibly

increases with age. The overall drop in the percentage of women with complete

54

information on date Of birth is about 12.4 percent from the youngest to the oldest cohort,
with the sharpest decreases among the two youngest cohorts.

Recall improves with 1-5 years of schooling such that on average about 59.3
percent of women in this schooling category reported complete dates of birth. About
72.9 percent of the youngest cohort 15-17 provided complete information on their date of
birth . This again drops sharply to 56 percent for the women in the 18-22 age. The
percentage stays around the upper 50's and then peaks at 65.4 percent for women in the
33-37 age group, thereafter it starts to drop to around the lower 50's for the next three
older cohorts. Small cell size for the oldest cohort 48-49 may account for the drop from
the 43-47 age group to the oldest cohort. Notice in particular, that the overall drop in
the percentage of women with complete information on date of birth for this schooling
group is about 18.4 percentage from the youngest to the oldest cohorts. This indicates a
bigger drop in the completeness of information for this schooling group over those with
no schooling.

Memory improves still further for women with six years of schooling such that
on the average above 69.0 percent of women with six years of schooling provided
complete information on their date of birth. In particular about 77.4 percent of the
youngest cohOrt 15-17 provided complete information on their date of birth, this drops
again sharply to 67.4 percent for the 18-22 age group. It then stays fairly constant
around the upper 60's and peaks at 74.2 percent for the 43-47 age group, then falls sharply
with the Oldest cohort ages 48-49. Small cell size for the Oldest cohort in this schooling

group may account for the sudden drop in the reporting of information regarding their

55

date of birth. The overall drop in the completeness of information from the youngest to
those in the 43-47 age group is about 3.2 percent and that from the youngest to the 48-49
age group is as high as 20.3 percent. This is in part accounted for by the huge drop from
the 43-47 age group to the 48-49 age group which is probably the result of the small cell
size mentioned above.

Recall improves dramatically for women with seven or more years of schooling
such that on the average about 91.7 percent of the women furnished complete information
on their date of birth. The percentage Of women with complete information on date of
birth increases faintly from 92.4 percent for women in the 15-17 age group to 92.9
percent for those in the 18-22 age group. It drops to 90.0 percent for the 23-27 age group
and remains fairly constant around 90 percent over the next three older cohorts It then
reaches its lowest point at 77.4 percent for the 43-47 age group and rises to 100 percent
for the 48-49 age. The small cell size for the two oldest cohorts in comparison with the
other cohorts may account for the unusually accurate reporting by these two oldest
cohorts. Ignoring the last two cohorts, we observe that the drop in the reporting of
information on the histories is about 1.2 percent form the youngest cohort to women in
the 38-42 age group, indicating a more gentle fall than women in the lower schooling
groups. It should also be mentioned that when one of the date items was missing in all
the schooling categories, it was more likely to be the month of birth as evident in column
3 of Table 4.2, but the year of birth was more likely to be consistent with the current age

of the respondent.

56

TABLE 4.2

Percent Distribution of All Women by Completeness of information on Date of Birth, by Age Group and
by Years of Schooling. Nigeria Demographic and Health Survey, 1990.

 

 

Month Month Year and Year and Age Total Sarnpl
and Year and Age Age Age Reported, e
Reported, Reported, Reported, Reported, Year & Size
NO Year Month Year Month
ALL WOMEN Imputation Imputed Imputed lgnored‘ Imputed
Age Group
15-17 75.4 -- 1.4 23.2 -- 100.0 1039
18-22 69.3 0.1 1.0 29.7 -- 100.0 1831
23-27 63 .7 - 1.8 34.6 -- 100.0 1684
28-32 55.0 -- 1.8 43.1 0.1 100.0 1587
33-37 52.8 - 5.3 41.9 - 100.0 955
38-42 44.3 -- 2.4 53.2 0.1 100.0 951
43-47 46.2 - 4.0 49.8 -- 100.0 572
48-49 46.3 - 1.9 51.9 -- 100.0 162
Total 59.9 0.0 2.2 37.9 0.0 100.0 8781
0 yrs of
Schooling
Age Group
15-17 53.9 -- 1.2 44.9 -- 100.0 321
18-22 44.9 0.3 1.0 53.8 -- 100.0 615
23-27 43.2 -- 1.6 55.2 -- 100.0 748
28-32 43.6 - 1.2 55.1 0.1 100.0 973
33-37 41.8 -- 4.2 54.0 -- 100.0 620
38-42 35.4 -- 2.5 62.0 0.1 100.0 693
43-47 38.9 -- 3.7 57.5 -- 100.0 435
48-49 41.5 -- 2.2 56.3 - 100.0 135
Total 42.4 0.0 2.1 55.6 0.0 100.0 4540
1-5 yrs of
Schooling
Age Group
15-17 72.9 -- 3.4 23.7 - 100.0 1 18
18-22 56.0 -- 2.2 41.8 - 100.0 134
23-27 57.7 -- 3.6 38.7 -- 100.0 1 l 1
28-32 53.0 -- 9.0 38.1 -- 100.0 134
33-37 65.4 -- 11.5 23.1 -- 100.0 104
38-42 52.3 - 7.0 40.7 -- 100.0 86
43-47 56.8 -- 9.1 34.1 - 100.0 44
48-49 54.5 -- -- 45.5 -- 100.0 1 1
Total 59.3 - 6.1 34.6 - 100.0 742

 

57

TABLE 4.2 continues

 

Month and Month Year and Year and Age Total Sam
Year and Age Age Age Reported, ple
Reported, Reported, Reported, Reported, Year & Size
No Year Month Year Month

Imputation Imputed Imputed lgnored“ Imputed

 

6 yrs of

Schooling

Age Group

15-17 77.4 -- 0.5 22.1 -- 100.0 199
18-22 67.4 - 2.0 30.5 - 100.0 347
23-27 69.0 -- 2.4 28.6 -- 100.0 294
28-32 66.2 -- 1.7 32.0 - 100.0 231
33-37 65.9 - 7.3 26.8 -- 100.0 123
38-42 66.0 -- -— 34.0 -- 100.0 103
43-47 74.2 -- 1.6 24.2 - 100.0 62
48-49 57.1 -- -- 42.9 -- 100.0 7
Total 69.0 — 2.1 28.8 - 100.0 1366
7+ yrs of

Schooling

Age Group

15-17 92.4 -- 1.5 6.0 -- 100.0 397
18-22 92.9 - 0.3 6.8 -- 100.0 733
23-27 90.0 -- 1.1 7.9 - 100.0 529
28-32 91.1 - 0.4 8.5 - 100.0 246
33-37 88.9 - 3.7 7.4 - 100.0 108
38-42 91.2 - - 8.8 - 100.0 68
43-47 77.4 -- 6.5 16.1 - 100.0 31
48-49 100.0 -- - -- -- 100.0 9
Total 91.7 - 1.0 7.4 - 100.0 2121

 

I'Year and Age was given but the year of birth plus the age add up to the year of interview so the years
was ignored.

58

Completeness of Information on Birth Histories of Respondents

We again make reference to Table D4 of the NDHS final report (columns 1, and
5-7), which presents information on children ever born with complete date of births.
According to Table D4 for children born in the year of the survey, about 97.9 percent had
complete information reported on their date of birth. This declines steadily as we move
further away from the year of the survey, for instance about 95.0 percent of children born
a year before the survey had complete information on their date of birth. In comparison
about 82.6 percent of children born four years before the survey had complete
information on their date of birth and only 75.4 percent of those born nine years before
the survey had complete date of birth.

Information on date of birth is more complete on children who are living than
those who are dead for each calendar year before the survey. For example, among
children born at the time of the survey and living, about 98.3 percent of them had
complete information on date of birth, however for those children who were dead about
91.7 percent of them had complete information. The gap widens for births that occur
several years before the interview. Notice in particular, for births that occurred five
years before the survey, we had about 87.7 percent complete date of birth reported for
those who were living and only 73.6 percent for children who were dead at the time of the
survey. The lower percentage reporting of date of birth for children who are dead is
consistent with either memory lapse or deliberate displacement of information on dead
children, or unwillingness to talk about dead children for cultural and religious reasons.

It should also be mentioned that complete information about the birthdate of a child could

59

also be inaccurate when parents are consistently displacing birth information which is

hard to detect.

Proportion of Deaths

The importance of accurate reporting of birth histories in demographic research
can not be overemphasized since most studies on fertility and mortality depend on this
information. A number of factors may account for errors arising from the reporting of
births, first as mentioned earlier, older women have more births to report than the
younger women, so the younger women can report with a greater degree of accuracy.
Memory lapse which is associated with old age becomes a problem for the older women.
Older women, who should have higher child and infant mortality rates, may have
difiiculty'remembering all of their live births, that subsequently died. If so, reported
mortality rates would be biased downwards. To this end, we decided to look at the
number of deaths per 1,000 live births to see if older women are reporting lower death
rates. This could account for the underreporting of births we encountered when looking
at the cumulative fertility at various attained ages.

Table 4.3 documents the proportion of deaths per 1,000 live births. We anticipate
that in the absence of misreporting of birth histories, older women will report higher
deaths than younger women, partly because of improvement in health and better nutrition
for children born to younger women than the older women and also because of a rise in

education with younger cohorts. For all women in the sample, the number of deaths

60

increases continuously fiom 116 with the 15-17 age group to 223 for women ages 43-47
and then drops slightly with women in the 48-49 age group.

Proportion of deaths per 1,000 live births for women with no schooling also
increases at first from 114 with the youngest age group to 190 for women in the 23-27
age group, then drops suddenly with the 28-32 age group. Thereafter it continues to rise
till the 4347 age group and then falls with the oldest cohort. There is no cell size
problems with the oldest women in either the all women or women with no schooling
distribution of deaths, so it is possible that the oldest women are underreporting the
number of dead live births. However, it could also be real reflecting some of the
macroeconomic shocks that have hit the Nigerian economy such as the Nigeria Civil War
(1967-70), the oil boom (1973-80) and subsequent oil crisis (1981-1986). As we can see
later in the analysis, the drop in the deaths per 1,000 live births for women in the 28-32
age group is also consistent with number of births peaking with this age group for the all
women sampled and women with no schooling.

We do not observe any definite pattern of deaths for women with 1-6 years and
seven or more years of schooling. Consequently, we do not have any convincing
evidence to support the fact that schooling improves the reporting of death rates since the

pattern is not conclusive for women with 1-6 or seven or more years of schooling.

6]

 

 

 

.2 9.2 3% Es 3 an. 8. a: .53.
a 2 as m: z. n: 5 :N 3.3.
.n 3. :4 8n 8 a: an RN :1?
S a: .3 n8 3. 9: EN 2: at:
3. am an 8a 3 2: 2” .5 2-3
am can So a: 8 m: a: o: 2.:
can a: 2: an. a N: 8. 3: 2-2
a: «8 5.. .3 z ”a a: a: «2:
2 a 2. 3 cm : z. 2. 2-2
955
&<
38.3 383 ”5.8.3 352% «5323 ”5.8.3 ”5.8.3 3.35
a 30> z a 3% 2 Co 50> c 4.2 no 50> +5 .8 30> 2 Ca 28> o
mum 3.5% 958:8 mo #53

 

.23. £32 seesaw Co as; 3 3:8 25 83 .2. 2:3: s .352 .52

m6 ”mu—ah.

62

Cumulative Fertility at Various Amined Ages

Table 4.4 documents the cumulative fertility at various attained ages. From the
birth histories we reconstructed cumulative fertility at ages 18, 23, 28, 33, 38 and 43.
With these variables we then calculated the mean number of children ever born at each
attained age for the different birth cohorts. This approach to the calculation of the
number of children ever born depends heavily on the accuracy of the birth histories and
accurate reporting of respondents’ own age or birthdate. All other things being equal, in
the absence of age misreporting or age forward telescoping by the respondents and that
of their children, one should expect that the older women should have more children at
each of the attained ages. This is because they have lower levels of schooling, and
higher infant mortality rates. It is also possible that the oil boom and subsequent oil
slump may have affected the younger and older women differently on the perception of
childbearing. However, we observe an inverted-U shaped profile for cumulative fertility
as we move from the oldest to the youngest women in the survey.

As Table 4.4 demonstrates, cumulative fertility at various attained ages first
increases steadily as we move from oldest to the youngest women. For instance,
cumulative fertility at age 18 rises steadily but gently from 0.31 with women ages 48-49
to 0.60 with the 28-32 age group and then declines over women in the next two younger
age groups. Cumulative fertility at age 23 also rises steadily, for example, women in the
48-49 age group on average have 1.41 mean CEB age 23, this rises uniformly and peaks
at 1.98 for women in the 28-32 age group and falls over the next youngest age group.

Cumulative fertility at age 28, however rises steadily from 2.83 and peaks later at 3.56

63

with the 33-37 age group and begins to fall with the next younger women. We observe a
continuous rise from the oldest to the youngest women for cumulative fertility at age 33.
A number of factors may account for the above distribution of cumulative births.
First it is much easier for the younger women to remember and report with accuracy the
number of children ever born because they have smaller numbers to report than the older
women. Second since age is involved here, memory lapse associated with old age
becomes more important. Third, younger women may have fewer child and infant deaths
than the older women, and since there may be the tendency for mothers to underreport
their dead children, underreporting of births would be greater for the older women than
the younger women in the survey. A fourth reason may be social and cultural norms and
taboos associated with bringing into memory of a dead child. Since older women may be
more culture preservers than the younger women, they may adhere more to social and
cultural norms and may underreport their dead children more than the younger women.
Another problem is age forward telescoping. This as mentioned earlier is the
forward displacement of events that happened in the distant past closer to the survey
period than they actually occurred. If older women are consistently displacing events
closer to the survey, then we should expect underreporting of births by older women at
younger ages and more accurate reporting at older ages. This may account for the initial
rising trend in cumulative fertility among the older cohorts and then the subsequent fall in
fertility among the younger cohorts we experience above. It may also explain why
cumulative fertility at age 23 peaks with the 28-32 age group, they are more likely to

remember with accuracy their cumulative fertility at age 23 than the older women before

64

them. This may also account for the fall in cumulative fertility at age 23 among the 23-
27 age group and possible sign of accurate reporting of birth by women in this age group.
This may also be a valid explanation for cumulative fertility at age 28 peaking with the
33-37 age group. However, our data cover only women in the 15—49 age group so we
have fewer cohorts to observe and compare cumulative fertility at older ages.

Another reason for the inverted U-shaped birth distribution may be real economic
shocks in the Nigerian economy, ranging from the Civil War, the oil boom and
subsequent oil slump. The Nigerian Civil War from May 1967 to January 1970 may have
delayed marriage for the older cohorts. This may have resulted in the late onset of first
births and subsequent fewer number of children ever born for the older cohorts. The oil
boom in the 1970's may have a differential impact on women in terms of their desire to
have children. Women in the 28-32 age group would have been 11-15 years in 1973-74
when the oil boom period started lasting into 1979-80, making them 17-21. This may
have made it more attractive for these young women to marry and have children than the
long term investment in schooling and subsequent delay of maniage. This may also be
true for cumulative fertility at age 28, which peaks with women in the 33-37 age group,
and may also explain the inverted U-shaped distribution of cumulative fertility at various

attained ages.

65

TABLE 4.4 - Cumulative Fertility At Various Attained Ages By Five-Year Age
Group. Nigeria, 1990.

 

 

 

 

Sample
M Size
18 23 28 33 38 43

ALL WOMEN

Age Group

1517 -- - .- - -- -- 1039
18-22 0.44 -- .- -- -- -- 1831
23-27 0.53 1.73 -- -- -- -- 1684
28-32 0.60 1.98 3.45 .- -- -- 1587
33-37 0.53 1.85 3.56 4.97 -- -- 955
38-42 0.52 ' 1.65 3.10 4.60 5.63 -- 951
43.47 0.48 1.63 2.98 4.41 5.71 6.51 572
48-49 0.31 1.41 2.83 4.19 5.33 6.14 162

 

Source: Calculated from Demographic and Health Survey, Nigeria, I990

66

Cumulative Fertility at Various Attained Ages by Years of Schooling

We anticipated that schooling may improve the memory of women and thus
improve the reporting of cumulative births at various attained ages. First the better
educated women may have fewer number of children and thus may be able to report with
greater accuracy than the less educated who have more births to report. Second since the
problem with memory lapse increases with age, we should see less accurate reporting
among the oldest women who are least educated than the better educated older women.

To investigate the influence of schooling on the memory of respondents, we
decided to look at cumulative fertility at various attained ages for women with different
levels of schooling. This we did over women with no schooling, 1-5 years of schooling,
six years of schooling and seven or more years of schooling, the results are presented in
Table 4.5. According to Table 4.5, cumulative fertility at the various attained ages for
women with no schooling increases uniformly as we move from the oldest to the
youngest cohort. For example, cumulative fertility at age 23 increases steadily from the
mean CEB of 1.47 with women ages 48-49 to 2.32 with those ages 23-27. This rising
trend is also observed for cumulative fertility at ages 28 and 33. However, for cumulative
fertility at age 38 we observe first an immediate rise from 5.37 with the oldest cohort to
5.62 with the next oldest cohort and then drops to 5.50 with the 3842 cohort. This may
be consistent with the problems of forward age telescoping, in that births at older ages are
closer to the survey period and may be remembered and reported with greater accuracy
than births that occurred at younger ages and thus in the distant past. However, since we

have only three cohorts to compare it is hard to draw any definite conclusions.

67

We do not see much of a pattern for cumulative fertility at age 18 for women with
1-5 years of schooling. Cumulative fertility at age 23 for women with 1-5 years of
schooling first falls slightly from 1.64 with the oldest cohort to 1.61 with the next older
cohort, thereafter it increases uniformly from the oldest to the youngest cohort reaching
its highest point at 2.23 with the 23-27 cohort. We observe a steady increase in
cumulative fertility at age 28 from the oldest cohort reaching its peak at 4.04 with the 33-
37 cohort and then declines over the next younger cohort. For cumulative births at ages
33 and beyond we observe a uniform increase in fertility from the oldest to the youngest
cohort.

As- we move to women with six years of schooling, we observe a more
pronounced inverted U-shaped distribution very similar to the aggregate pattern of the
distribution of cumulative birth discussed in the previous section above. We do not see
much of a pattern for cumulative fertility at age 18. However, cumulative fertility at age
23 increases steadily from 0.57 with the oldest cohort, reaches its maximum at 1.90 with
the 33-37 cohort and then falls over the next two younger cohorts. Cumulative fertility at
age 28 also reaches its peak at 3.75 with the 33-37 cohort and then falls with the next
younger cohort. We observe a steady increases from the oldest to the youngest cohort for
cumulative fertility at age 33. For cumulative fertility at age 38 we observe first a rise
from 5.43 with the oldest cohort to 6.08 with the next older cohort and then falls to 5.92
with the 38-42 cohort. This is very similar to the pattern for cumulative fertility at age 38
for women with no schooling. As mentioned earlier this may be evidence to support age

forward telescoping but again the number of cohorts to compare is very few.

68

The pattern of births for women with seven or more years of schooling is
remarkably different from the other schooling categories discusses above. We do not see
any definite pattern for cumulative fertility at age 18. Cumulative fertility at ages 23 and
beyond, we see first an increase in fertility from the oldest cohort to the next older cohort
and thereafter a steady decline in fertility from the oldest to the youngest cohort. The
relatively small cell size for the oldest cohort for this schooling group may account for
the initial rise in cumulative fertility. However, it is also possible that the oldest cohort is
misreporting the number of births, an observation that is consistent with the reporting of
births by the oldest cohorts with the no schooling, primary or some primary witnessed
above. The distribution of cumulative fertility at various attained ages is very similar to
the distribution of the percentage of "ever-married" at various attained ages that we
discuss in chapter 4 of this study. This is very consistent with the data since first
marriages and births are closely related.

In looking at the cumulative fertilities for the different schooling categories, we
make several observations. First for women with no schooling we observe a steady
increases in fertility from the oldest to the youngest cohort. This may be consistent with
misreporting of births at various attained ages as a result of memory lapse. For
cumulative fertility at age 38 we begin to see some slight evidence to support forward age
telescoping, older people reporting better at older ages

We continue to see a rising trend in fertility from the oldest to the youngest cohort
for women with 1-5 years of schooling for crunulative fertility at various attained ages,

with the exception of cumulative fertility at age 28. At age 28 the pattern is more of an

69

inverted U-shape in form, it rises initially reaches its peak and then falls. This seems to
suggest that women with 1-5 years of schooling may not be any different from women
with no schooling in fertility decision making.

The cumulative fertility for women with six years of schooling exhibits an
inverted U-shaped distribution and this is observed for cumulative fertility at ages 23 and
beyond. However, for women with seven or more years of schooling we see a steady
decline in fertility from the youngest to the oldest cohort.

Many factors may account for the observed distribution of births among women
of different educational status. It is possible that women with no schooling, 1-5 years and
six years of schooling are misreporting their age at each of their births, this being the case
especially for the older women. Memory lapse associated with age may be a factor in the
response given by women with no schooling, primary and some primary education. This
is more likely since reporting of births improves drastically among women with seven or
more years of schooling. The distribution of cumulative births among women with seven
or more years of schooling may be real and may not reflect memory lapse or forward age
telescoping. This is consistent with seven or more years of schooling improving
drastically the completeness of reporting of respondents age discussed earlier.

Another factor may be age telescoping, the older women displacing their births
forward and closer to the survey period than they actually occurred. This may explain the
pattern of cumulative births at age 38 for women with no schooling and women with six
years of schooling. We do not have enough evidence to support this argument since we

have very few cohorts to compare at cumulative fertility at higher attained ages.

70

TABLE 4.5. Cumulative Fertility At Various Attained Ages By Five-Year Age Group
and by Years of Schooling, Nigeria, 1990

 

 

 

 

 

 

 

By Age Sample
Size
18 23 28 33 38 43
0 years of Schoolirg
Age Group
1 5-17 -- -- -~ -- -- -- 32 1
18-22 0.84 - - - - -- 615
23-27 0.87 2.32 -- -- -- -- 748
28-32 0.77 2.26 3.71 - - -- 973
33-37 0.61 1.92 3.59 4.97 - -- 620
38-42 0.58 1.67 3.03 4.49 5.50 -- 693
43-47 0.54 1.65 2.97 4.34 5.62 6.48 435
48-49 0.33 1.47 2.90 4.25 5.37 6.16 135
1-5 years of Schoolirg
Age group
15-17 -- - -- -- - -- 1 18
18-22 0.54 -- -- - -- -- 134
23-27 0.66 2.23 -- -- -- -- 1 1 1
28-32 0.53 2.13 3.90 - -- -- 134
33-37 0.53 2.13 4.04 5.70 -- -- 104
38—42 0.45 1.77 3.62 5.50 6.87 - 86
4347 0.27 1.61 3.16 4.93 6.23 6.89 44
48-49 0.55 1.64 2.64 3.55 4.73 5.73 11
6 years of Schooling
Age Group
15-17 - -- -- -- -- - 199
18-22 0.42 —- - -- - - 347
23-27 0.40 1 .67 -- -- - - 294
28-32 0.43 1 .82 3.43 -- - -- 23 1
33-37 0.43 1.90 3.75 5.25 - -- 123
38-42 0.32 1.75 3.44 4.95 5.92 - 103
43-47 0.32 1.65 2.90 4.50 6.08 6.68 62
48-49 0.00 0.57 2.71 4.00 5.43 6.29 7
7+ years of Schoolng
Age group
15-17 -- -- -- -- -- -- 397
18—22 0.09 -- -- -- -- - 733
23-27 0.10 0.82 -- -- -- -- 529
28-32 0.15 0.96 2.23 -- -- -- 246
33-37 0.19 1.14 2.71 3.93 - - 108
38-42 0.26 1.24 2.74 4.03 4.93 - 68
43-47 0.19 1.32 3.03 4.45 5.55 6.06 31
48-49 0.00 0.89 2.22 4.22 5.44 6.22 9

 

71

Some of the fertility patterns could also be real resulting from macroeconomic shocks
namely, the Nigeria Civil War from May 1967 to January 1970, the oil boom in the 19705
and subsequent oil crisis in the 1980s. The oil boom may have made it attractive for
women especially the least educated to marry and have children. In addition credit
constraints and increasing bride price may have played a role in delaying marriages and
subsequent onset of births. The oil boom, and the credit constraints and bride price may
have affected women with different educational levels differently, which may explain the
inverted U-shaped distribution that we encounter, in particular, for women with six years
of schooling. The Nigeria Civil War may also have delayed marriage and onset of births
for the older cohort.

In what follows we investigate in some details whether the Nigerian Civil war had

any effect on fertility among the older cohorts.

Is There a Baby Boom Effect?

There is the tendency for fertility to decline during a war period and to surge
years immediately after a war. This is partly because men are normally taken to war
whilst the wives wait and also possible famine that may be associated with war periods.
We decided to look at the mean CEB for women in the age group 43-49 around the
Nigeria Civil War to see if there is any sign of war effect on the reporting of births. If
there is war effect, which group of women experienced the greatest impact of the war,
and if that could be the reason for subsequent underreporting of births. If there is no war

effect could be a memory lapse or real macroeconomic shocks?

72

In Table 4.6 we document the mean number of children ever born three years
(1964-66) before the Nigerian Civil War, three years (1968-70) during the Civil War and
three years (1971-73) after the Civil War for women in the 43-49 birth cohort.‘5 As we
move horizontally following each cohort of women for the three sub-periods around the
Civil War, we notice that for the all women sampled, the mean CEB first rises from 0.70
three years preceding the war to 0.83 three years during the war and then drops to 0.79
three years after the war for women ages 43-47 in 1990. Mean CEB for the oldest cohort
48-49 remains constant at 0.85 for the three years preceding the war and three years
during the war and then drops to 0.75 three years after the war.

In both of the age groups we see a drop in fertility after the war which could be
either misreporting of the number of births or could be real economic shocks or
systematic displacement of births by these oldest cohort which is hard to distinguish.

We do not observe any surprises in the reporting of mean CEB as we move vertically
among the age groups for the three subperiods around the war under study.

We decided to group women into years of schooling to investigate whether years
of schooling improves the memory of women for births that occurred 17-23 years before
the survey. For women with no schooling similar patterns are observed as we did with
the overall women. Women ages 43-47 in 1990 experience a rise in fertility from the
years before the Civil War to the years during the Civil War and then fertility drops three

years after the Civil War. We observe a continuous fall in fertility for women in the 48-

 

‘ The Nigerian Civil War was from May 1967 to January 1970. Because of the time it takes between
conception and birth we decided to look at three years before the war , three years into the war and three
years after the Civil War.

73

49 age group. For similar reasons given above, the two oldest cohorts 43-47 and 48-49
may be misreporting their mean CEB three years afier the Civil War.

We begin to see a little bit of a baby boom effect for the fertility pattern of women
with 1-5 years and six years of schooling. For example, women in the 43 -47 age group,
we begin to see a little bit of baby boom effect, in that fertility drops during the war
period and then rises again after the war. Notice that for women with 1-5 years of
schooling who were 43—47 in 1990, mean CEB drops from 0.91 before to the war to 0.80
during the war and then rises to 1.14 afier the war. For women in the 48-49 age group
we observe a continuous fall in fertility for the three subperiods, whereas for women with
six years of schooling we observe a rise and then a fall in fertility over the three periods.
The observed pattern for the oldest women in these two schooling category may be
attributable to the small cell size problem or may be misreporting of mean CEB or from
real macroeconomic shocks.

The observed pattern of fertility before, during and after the Civil War for the
women with seven or years of schooling also rises during the war period and then falls
after the war for women in the 43-49 birth cohorts. Again small cell size may account
for the drop in fertility after the war for these birth cohorts or pure misreporting of births
by women in this age group.

As we move vertically between the different age groups and the different
schooling categories, again we do not see any surprises in the pattern of mean CEB for
women with no schooling, six years of schooling and seven or more years of schooling.

We see a sudden drop of mean CEB from women in the 43-47 to the 48-49 age groups for

74

women with 1-5 years of schooling. It could be either underreporting of births by the
oldest cohort or the small cell size for the oldest cohort with 1-5 years of schooling.

From the analysis of children ever born before, during and after the Nigerian
Civil War, we do not see evidence to support a baby boom effect. Since these three
periods span a period of 17-23 years before the survey, one possible reason may be birth
misreporting by these older women. We could not do comparison for the younger cohort
since they were too young to have children over the period under study. We have already
established the fact that memory lapse is greatest among the older cohorts in the sample,
so misreporting of the number of births may be an important factor underlying the pattern

of births we observe in our Civil War analysis.

75

TABLE 4.6 - Mean Number of Children Ever Born Three Years Prior, During and After

 

 

 

 

 

 

 

the Nigerian Civil War

CEB 1964-66 CEB 1968-70 CEB 1971-73 Sample Size
ALL WOMEN (43-49)
Age in 1990
43-47 0.70 0.83 0.79 572
48-49 0.85 0.85 0.75 162
0 years of Schooling
Age in 1990
43-47 0.67 0.83 0.73 435
48-49 0.87 0.81 0.76 135
1-5 years of Schooling
Age in 1990
43-47 0.91 0.80 1.14 44
48-49 ' 0.64 0.55 0.36 11
6 years of Schoolig
Age in 1990
43-47 0.82 0.73 0.84 62
48-49 0.86 1.43 0.57 7
7+ years of Schooling
Age in 1990
43-47 0.55 1.19 0.97 31
48-49 0.89 1 .22 1.1 1

 

Source: Calculated from Demographic and Health Survey, N igeria, l 990.

76

4.4. SUMMARY AND CONCLUSIONS

In this chapter we have examined some of the potential measurement errors in
the NDHS, 1990, with particular reference to the age reporting by respondents, reporting
of birth histories, age at first marriage and age at fust birth. We found evidence to
support age heaping. The Whipple's Index of age concentration on digit “0" and “5" was
224.6 which is very rough by the United Nations standards. This suggests that any
estimates based on age grouping should be centered around the preferred digits to avoid
serious bias estimates. We also found some slight evidence on age heaping for age at
first birth and age at first marriage, but this is not as big a problem as the age heaping
with the respondents age. The Whipple's index for age concentration was 121.73 and
112.78 for age at first marriage and age at first birth respectively, which is approximate
by United Nations standards. '

There was also evidence of problems associated with memory lapse with the
oldest women and also women with no schooling or some primary education. Older
women may be consistently underestimating their number of children ever born at
various attained ages. For example, using the birth histories, we reconstructed
retrospective cumulative fertilities at various attained ages, and for all the women
sample, the older women consistently reported lower cumulative births at each of the
attained ages than the younger women. Similar observations were made for women with
no schooling, 1-5 years of schooling and six years of schooling. For women with six
years of schooling,mean CEB eventually falls for the younger cohorts but the

distribution first rises reaches its peak and then falls. Seven or more years of schooling

77

improve the memory of the respondents greatly even though the problem of misreporting
of births for the oldest group 48-49 still remains. Small cell size of women ages 48-49 in
this schooling group may account for the apparent underreporting of births.

In looking at the number of deaths per 1,000 live births, we also noticed that
deaths drop suddenly from the 43-47 age group to the 48-49 age group for all women in
the sample and for women with no schooling. This may be an indication of under-
reporting of deaths by the oldest women, since chances are that earlier births by these
women may have resulted in deaths. Moreover, the older women may have more recalls
to do than the youngest women because of their longer exposure to childbearing.

We also anticipated that the Nigerian Civil War may have affected fertility in the
years before the war, during the war and after the war. However, the war analysis
restricts us to using the women ages 43-49 in 1990 since the analysis period is 17-23
years before the survey. We did not see any baby boom effect, this could be under
estimation of the number of births by the women in this age groups because of the
memory recalls involved or could be real shocks.

In our later analysis we decided to center ages around the digits "0" and "5" to
correct for the problems with age heaping. We had done our analysis with the traditional
age groupings 15-19, 20-24. 24-29 etc., and found our fertility analysis to be misleading
so we decided to use the age grouping 18-22, 23-27, 28-32, etc., dropping the 15-17 and
48-49 to give equal probability of occurrence for the digits "0" and "5". We saw a
significant improvement in our analysis which are presented in chapters 4 and 5 of the

present study.

78

Demographic surveys have improved recently with more advanced data collection
procedures. DHS data collection has been assessed by its staff and found the quality
satisfactory. A comparison of DHS data with that of WFS by DHS staff often revealed
a more complete reporting of ages by women in the DHS than in the WFS. Our analysis
in this present chapter has been restricted however only to the NDHS, a more complete
study is to do an independent comparative study of the NDHS and that of the Nigeria
WFS, but this is beyond the scope of the present study. However, the evidence of age
heaping especially that of the respondent age is overwhelming and we will recommend
that correction for age heaping be done especially with fertility and mortality estimates

which are based on age groupings

79

CHAPTER 5

FERTILITY TRENDS IN NIGERIA

5.1. INTRODUCTION

 

Since 1965 most parts of the world have experienced a fertility transition. Sub-
Saharan Africa is yet to have undergone any major decline in fertility (National Research
Council, 1993). Botswana, Kenya and Zimbabwe are often cited as the three sub-
Saharan countries in Africa that have experienced an irreversible decline in fertility (van
de Walle and Foster, 1990; World Bank, 1990). However, its extent and permanency is
ofien debated in the empirical literature (Caldwell, 1992; Rutenberg and Diamond, 1993;
Thomas and Muvanda, 1994)

Recently however, with the availability of data from countries participating in the
Demographic and Health Surveys and more sophisticated fertility analysis, there is some
evidence to support the idea of a fertility transition in some countries and regions in Sub-
Saharan African countries besides Botswana, Kenya and Zimbabwe, notable among
which is the Southwest and Southeast regions of Nigeria (National Research Council,
1993). Recent Nigeria Demographic and Health Survey data indicate that there has been
a drop of 1.3 in the total fertility rate, within the period eight years and four years before
the survey (Cohen, 1993). By 1991, there was evidence to suggest that there has been
over 10 percent drop in fertility in the Southwest Nigeria and possibly in Southeast

Nigeria (N igeria/DHS, 1991; Reinis, Rutstein and Ajayi, 1991).

80

This chapter analyzes fertility behavior of women in Nigeria, with particular
reference to the three regions, namely, the Southwest, the Southeast and the North. The
chapter starts out first by looking at the trend in female schooling. Second, we examine
fertility trends in Nigeria as a whole by looking at fertility correlates such as cumulative
fertility at various attained ages, first births and first marriages at various attained ages.
Finally we also investigate family planning policies and the use of contraceptives among
women of childbearing ages.

The analysis of the data establishes the fact that female schooling has increased
greatly among the youngest cohorts, particularly those living in the south. Women living
in Lagos, other areas in the southwest and in the southeast are better educated than those
living in the rural and northern areas.

A look at fertility in general does not reveal any decisive decline in fertility. For
example, the children ever born declines rapidly from the oldest to the youngest women
but this may be because the older women have had a longer exposure to pregnancy and
childbearing than the younger women. Cumulative fertility at various attained ages for
women actually increases initially, reaches a peak and then eventually falls with the
youngest cohorts as we progress from the oldest to the youngest cohorts. Whilst this is
true among all the women sampled and also in the southwest and the southeast, among
women in the north, even though cumulative fertility at various attained ages eventually
falls, sometimes that of the youngest women is actually higher than that of the oldest

women (see Figure 5.2). We make similar observations with births and marital patterns.

81

Contraceptive use which was only one percent among currently married women
has since 1989 risen to 6 percent (3.5 modern and 2.5 traditional methods). Among all
women about 45.7 percent know about a contraceptive method and for the currently
married about 31.9 percent know a contraceptive method. About 29.1 percent of women
with no schooling know a contraceptive method, this compares with about 90.8 percent
of women who have completed secondary school or higher. About 9.6 percent of all
women live under one mile to a facility where family planning services are provided.

The rest of the chapter is organized as follows. Section II examines the trends in
female schooling over recent years. Section III discusses the fertility trends in Nigeria
without any reference to female schooling. Section IV investigates family planning
policies and the knowledge and use of contraceptives among all women and currently

married women. Section V concludes the chapter.

5.2. TRENDS IN FEMALE SCHOOLING

 

In Figures 5.1A through 5.1H, we depict the trend in female schooling in Nigeria
between 1943 and 1972 based on the NDHS, 1990 data. According to Figure 5.1A there
has been a steady increase in the mean level of completed years of schooling for women
in Nigeria from the oldest to the youngest birth cohort. The average mean level of
completed schooling for women born in the 1943-47 is 1.47, in comparison, women born
in 1968-72 have completed 5.54 mean years of schooling. The highest increases in
schooling was witnessed by women in the 1963-57 birth cohort compared to the 1958-62

cohort.

82

As depicted in Figure 5. 13, in addition to the rise in female educational
attainments, the proportion of women who have never received any formal education has
dropped drastically from 76.1 percent to 33.6 percent from the oldest to the youngest
birth cohorts. The percentage of women who complete primary schooling and beyond
has also risen substantially from about 16.3 percent with women born in 1943-47 and
more than tripling to 59.1 percent with those born in 1968-72.

In looking at the educational attainment for women in the area of residence, we
observe that women residing in Lagos and other urban areas are better educated than
women in the rural areas an observation we make for all birth cohorts (see Figure 5. 1 C)
In the regional samples, we observe that women in Lagos and other areas in the
Southwest, and Southeast to a lesser extent, on average have higher mean years of
completed schooling than w0men in the North (Figure 5.1D).

A possible reason for the high concentration of better educated women living in
Lagos and other Southwest might be the formal labor market opportunities that urban
living provides. In addition higher standard of living, more educated parents, and
availability of more and better quality schools in the urban settings may account for the
higher educational attainment of women residing in the urban areas (Behrman and

Deolalikar, 1988).

83

 

, /
/

MourYoaraotSehoollng
. 0|

 

 

 

1

1943-47 1048-62 1053-67 rose-ea toes-c1 tau-72
Birth Cohort

 

 

+Yra. ot Sch.

 

 

Figure 5.1A - Mean Years of Schooling. Female Cohorts Born
1943-72

 

A

\\

a

 

 

 

 

 

 

Pct. Distribution of Schooling
8

 

 

 

a
m
to
o. . . T T ,
tats-47 tats-a 153-a 1M m tau-12
BirthOohort
+Pct0yrs +Pc16yrs

 

 

 

Figure 5.1B- Percentage Distribution of Years of Schooling.
Female Cohorts Born 1943-72

84

momsmum 1

ManaotSchoollng

NUQOONCO

‘

 

 

 

 

 

Figure 5.1C & 5.1D- Mean Years of Schooling.
Female Cohorts Born 1943-72

 

 

 

0
10
'° 0
70 § 0
Q N
8
O
U £ a
0 3
0
fl 3
a) m
to E .6
o 0
m m m m m “n ”a no. we .. m ’H
“w “w
L+Lm+mw+fl Lttm+m+arws~+ml

 

 

Figure 5.1E & 5.1F- Percentage of Women with Zero Years of Schooling
Female Cohort Born 1943-72

85

 

33 7=="
,0 ‘\ / A

80 M

i: ‘7/‘//2:_
:: M;

 

 

 

 

 

 

 

PctauthutlonotYoanotSehoollng

 

 

 

10
o I I U I U I
1 943-47 1946-62 1 953-67 1956-62 1 963-67 1966-72
B lrth Cohort

 

[-o-Lagoa +Othor Urban +Rural I

 

Figure 5. lG- Percentage with 6+ Years of Schooling
Female Cohorts Born 1943-72

100
99
60
79
60
60
40
90
20
19

0
1943-47 1946-52 1953-57 . 1956-62 1963-67 1966-72

Blrth Cohort

MWotYnolmng

 

 

L -O—Lagoa +8outhoaat —o—Othor 8W +North ]

 

Figure 5.1H- Percentage with 6+ Years of Schooling
Female Cohorts Born 1943-72

86

5.3. FERTILITY TRENDS IN NIGERIA
Number of Children Ever Born.

In Table 5.1 we document the "number of children ever born" as a measure of
cumulative fertility. This includes all live births in a woman's life time, we should.
however, take note of the fact that the younger women in the survey have not completed
their fertility. Consequently, the mean number of children ever born (CEB) increases
rapidly as we move from the youngest to the oldest women. By the end of her
reproductive years a woman aged 43-47 would have had on the average 6.70 children.

On average for each age group, women in Lagos have the least number of
children, followed by those in the other urban areas. Women ages 43-47 residing in
Lagos on average have about 5.27 children at the end of their reproductive cycle. In
comparison, women in the same age group in the other areas of Southwest have 7.14,
those in the Southeast have 7.18, those in the rural areas have 7.02 and those in the North
have 6.45. Younger women ages 18-37 in the North report higher number of mean CEB
than younger women residing in either the Southeast or other areas in the Southwest. In
contrast, older women ages 38-47 in the North report lower mean CEB than their
counterparts living in the Southeast or other areas in the Southwest. In comparing the
Southeast and other areas in the Southwest, we observe that younger women ages 18-37
in the other areas of Southwest report fewer mean CEB than women of the same age
group in the Southeast, the reverse is true for women in the 38-47 age group. Women in

Lagos have smaller family size than women in the other residential areas.

87

TABLE 5.1
Distribution of All Women by Number of Children Ever Born (CEB) and Mean Number of
Living Children by Five-Year Age Group. Nigeria, 1990

 

 

 

 

v. with 111111366 Number Mean Mean no.
W of no. of of Living
Women CEB Children
, 0 1-3 4-6 7.9 10+ Total
ALL WOMEN
08-47)
Age Group
18-22 50.2 46.3 3.3 0.1 0.0 100.0 1831 0.91 0.76
23-27 17.5 56.2 25.0 1.1 0.1 100.0 1684 2.41 1.99
28-32 6.2 33.5 46.6 13.4 0.4 100.0 1587 4.06 3.45
33-37 3.0 17.5 44.3 30.8 4.4 100.0 955 5.46 4.40
38-42 4.8 16.7 33.6 33.6 11.1 100.0 951 5.93 4.56
43-47 3.8 15.4 24.7 36.5 19.6 100.0 572 6.70 4.97
Total 18.6 36.1 27.8 13.9 3.5 100.0 7580 3.55 2.82
LAGOS
Age Group
18-22 75.4 23.7 0.9 0.0 0.0 100.0 333 0.39 0.36
23-27 31.9 58.3 9.8 0.0 0.0 100.0 307 1.55 1.42
28-32 8.9 43.7 42.1 5.3 0.0 100.0 247 3.37 3.02
33-37 0.0 20.0 55.7 23.6 0.7 100.0 140 5.09 4.55
38-42 0.9 24.1 50.9 21.6 2.6 100.0 1 16 5.06 4.41
43-47 3.9 21.6 45.1 25.5 3.9 100.0 51 5.27 4.76
T0181 31.3 36.3 24.9 7.0 0.5 100.0 I 194 2.52 2.26
OTHER URBAN
Age Group
18-22 58.3 40.2 1 .4 0.2 0.0 100.0 503 0.69 0.60
23-27 ' 18.0 56.9 23.8 1 .3 0.0 100.0 399 2.34 2.05
28-32 6.4 30.5 50.8 12.3 0.0 100.0 390 4.05 3.52
33-37 ’ 3.8 17.3 44.9 30.2 4.0 100.0 225 5.36 4.54
38-42 ‘ 4.7 12.7 38.0 36.2 8.5 100.0 213 5.98 4.96
43-47 2.5 20.0 28.3 30.0 19.2 100.0 120 6.27 5.04
151.1 22.2 345 27.9 12.7 2.7 100.0 1850 3.29 2.79
RURAL
Age Group
18-22 37.8 57.0 5.1 0.1 0.0 100.0 995 1.19 0.97
23-27 12.8 55.3 30.3 1.4 0.2 100.0 978 2.71 2.14
28-32 5.4 32.0 46.0 15.9 0.7 100.0 950 4.25 3.36
33-37 3.6 16.9 28.9 35.0 13.7 100.0 590 5.59 4.31
38-42 5.6 16.7 28.9 35.0 13.7 100.0 622 6.09 4.45
43-47 4.2 13.2 20.9 39.9 21.7 100.0 401 7.02 4.97

 

88

TABLE 5.1 continues

 

 

 

 

% with Indicated Number Mean Mean no.
W of no. of of Living
J Women CEB Children
0 1-3 4-6 7-9 10+ Total
SOUTHEAST
Age Group
18-22 60.8 36.3 2.7 0.2 0.0 100.0 487 0.70 0.62
23-27 21.8 49.6 26.9 1.2 0.5 100.0 427 2.39 2.07
28-32 6.6 30.1 47.7 15.1 0.5 100.0 392 4.20 3.59
33-37 3.3 17.8 37.8 36.7 4.4 100.0 275 5.68 4.81
38-42 1.6 10.4 31.9 44.2 12.0 100.0 169 6.75 5.33
43-47 1.2 9.5 28.4 39.6 21.3 100.0 169 7.18 5.65
Total 21.5 29.9 27.3 17.2 4.1 100.0 2001 3.79 3.10
OTHER
SOUTHWEST
Age Group
18-22 66.0 33.0 0.7 0.0 0.0 100.0 276 0.48 0.42
23-27 21.7 66.0 12.3 0.0 0.0 100.0 212 1.83 1.58
28-32 3.0 36.9 48.5 11.1 0.5 100.0 198 3.99 3.39
33-37 0.0 10.7 56.7 28.7 4.0 100.0 150 5.73 4.70
38-42 0.6 7.4 44.6 36.6 10.9 100.0 175 6.39 5.05
43-47 0.8 10.9 26.9 42.9 18.5 100.0 119 7.14 5.53
Total 21.0 30.6 28.2 15.9 4.2 100.0 I 130 3.66 2.98
NORTH
Age Group
18-22 25.9 68.2 5.9 0.1 0.0 100.0 735 1.44 1.16
23-27 7.9 56.4 33 .9 1.9 0.0 100.0 738 2.95 2.29
28-32 5.9 30.9 46.9 15.7 0.5 100.0 750 4.24 3.31
33-37 5.1 19.0 40.0 30.0 5.9 100.0 390 5.35 3.93
38-42 9.8 22.5 25.2 29.3 13.2 100.0 409 5.51 3.92
43-47 7.3 20.6 ' 16.3 35.5 22.3 100.0 233 6.45 4.23
Total 1 1.3 41.9 28.9 13.8 4.1 100.0 3266 3.77 2.81

 

Source: Calculated from Demographic and Health Survey,. Nigeria, 1 990

89

Cumulative fertility: Another Look.

As mentioned earlier in our literature review chapter, the use of the total number
of children ever born as a measure of cumulative fertility has censoring problems
differentially for different age groups. The younger women in the sample have not
completed their fertility. Some studies get around this by using the most recent fertility,
usually children born five years before the survey. This still does not avoid the problem
of censorship. The youngest women in the survey in particular the 15-19 would be 10-
14 five years before the survey and may be too young to have any number of children
five years prior to the survey. In addition the two oldest age groups of women (40-44 and
45-49) would be 35-39 and 40-44 respectively five years prior to the survey and at the
end of their reproductive years and may not be having considerable number of births
within the five years prior to the survey. In this section an alternative measure of
cumulative fertility is provided. Using the retrospective birth histories we reconstructed
cumulative fertility for each woman at age 18, 23, 28, 33, 38 and 43. The main
drawback with this method is that it relies heavily on the accuracy of reporting of
mother's own age and that of each child born alive, however, it avoids the problem of
differential censoring discussed above. The breakdown of Southwest into Lagos and
other Southwest revealed very similar results so we decided to combine Lagos and other
Southwest into Southwest. We decided not to discuss cumulative fertility for the rural
and urban residents, we will look at this in some detail when we run the regressions in the

next chapter.

Figures 5.2A through 5.2D depict cumulative fertility at various attained ages for
the all women sample and the regional samples. As Figure 5.2A shows, among all the
women, the profile representing cumulative fertility at age 18 is very uniform. Starting
with a mean CEB of about 0.48 with women ages 43-47, it rises continuously but faintly
such that the next three younger cohorts have actually born more children than the oldest
cohort. It peaks at 0.60 with women ages 28-32 before it declines over the next two
younger age groups. The profile for cumulative fertility at age 23 on the other hand rises
faintly at first from 1.63 and then sharply reaching its peak at 1.98 with the 28-32 age
group before it declines. We make similar observations with cumulative fertility at age
28 peaking earlier with the 33-37 age group. We experience a steady increase in
cumulative fertility at age 33 from the oldest to the youngest women.

We make similar observations with cumulative fertility in the regions. In the
Southeast as Figure 5.2B shows, cumulative fertility at ages 18 and 23 fluctuates among
women ages 28-47 but declines over women ages 18-27 for cumulative fertility at age 18
and for women ages 23-27 for that of cumulative fertility at age 23. On the other hand,
that of cumulative fertility at ages 28 rises steadily fiom 3.56 with the oldest women and
reaches its highest point at 3.74 with the 33-37 age group before it declines with the next
younger age group. Cumulative fertility at age 33 on the other hand rises and then
declines from oldest to the younger women.

A look at cumulative fertility at various attained ages for women residing in the
Southwest also reveals an eventual fall in cumulative fertility at ages 18, 23 and 28. As

evident in Figure 5.2C, cumulative fertility at ages 18, 23 and 28 all rise initially reach a

91

peak and then decline. Cumulative fertility at age 33 on the other hand rises steadily
from the oldest to the youngest women.

Figure 5.2D shows the cumulative fertility at various attained ages for women
residing in the North. In the North cumulative fertility at various attained ages rise
uniformly fi‘om the oldest to the youngest women. That at ages 18 and 23 falls faintly
after reaching their highest points with women in the 23-27 and 28-32 age groups
respectively but still the cumulative fertility of the youngest cohort is higher than that of
the oldest women.

Overall there is little indication of a decline in cumulative fertility among all the
women at ages 18 and 23. At higher ages cumulative fertility of the youngest women are
even higher than the older women in the survey. This is possible cyclically but not a
long-run trend, although the recent claim of a fertility decline could be trend but we do
not know yet as these women are too young. It is also possible that the higher cumulative
fertility for the younger women by ages 28 or 33 signals changes in the timing of births.
This is possible if younger women delay their births but space them closely together so
that at the end of their reproductive years they have either the same or more children than
the older women. Empirical work on the life-cycle of women suggest that the more
educated women take time off and bunch their children together so that they may return
to work sooner.

At younger ages women in the Southwest report lower cumulative fertility than
their counterparts in the Southeast. At ages 18 and 23 women in the North report the

highest cumulative fertility. However, for cumulative fertility at ages 28 and 33 the

92

oldest women ages 38-47 in the North report fewer number of children than those in the
Southwest or the Southeast. This may be a possible sign of underreporting of children by
the oldest women in the North. This may also be real in that the North is the poorest
region in Nigeria and poor nutrition and health conditions may have contributed to low
fertility in this region.

We see some signs of a declining fertility among the youngest women in the
Southeast and Southwest for cumulative fertility at ages 18 and 23. For example,
cumulative fertility at ages 18 declines by about 46.7 percent over the two youngest age
groups and that at ages 23 declines by about 16 percent with the 23-27 age group for
women in the Southeast. Similarly, for women in the Southwest cumulative fertility at
age 18 declines by 64.7 percent over the two youngest age groups of women and that of
cumulative fertility at age 23 declines by 29.4 percent with the 23-27 age group.
However, we observe an increase in cumulative fertility in the two regions for cumulative
fertility at ages 28 and over, again we have different cohorts and this may be cyclical
effects or changes in the timing of births which is hard to distinguish. There is, however,
no indication of a fertility decline in the North.

One may conclude that a look at cumulative fertility does not suggest any decisive
decline in fertility among all the women and even the drop in either the Southwest or
Southeast is not conclusive. At younger ages yes, but one needs to follow and examine
the fertility patterns of the younger women for a longer period to see if it will remain at

the present level or there would be a change in their present fertility patterns.

93

Cumulative Fertility

Cumulative Fertility

0

u .

/

 

 

 

 

A A
- -

 

 

 

43-47 38-42 33-37 28-32 23-27 18-22

Age Group

 

[+cumfl8 +eumf23 +cum128 +cumffi

Figure 5.2A- Cumulative Fertility at various
Attained Ages. All Women, Nigeria, 1990

 

>

 

1

 

 

 

 

43-47 38-42 33-37 28-32 23-27 18-22

Age Group

 

[+cumf18 +cumf23 —-—cumf28 +cumf31fl

Figure 5.28- Cumulative Fertility at various
Attained Ages. Southeast, Nigeria, 1990.

94

Cumulative Fertility

 

 

 

 

 

A J

A - - A
' - A
- -
1' I U I 1

43-47 38-42 33-37 28-32 23-27 1 8-22

Cumulative Fertility

\

0

Age Group

 

 

+ cumfl8 + cumf23 -°- cumf28 + cumf33

 

 

Figure 5.2C- Cumulative Fertility at various
Attained Ages. Southwest, Nigeria, 1990.

 

 

 

 

Ir

-
‘ Q
-

 

 

1 T I T I

43-47 38-42 33-37 28-32 23-27 18-22

Age Group

 

+ cumft 8 —o— cumf23 —-— cumf28 + cumf33

 

Figure 5.2D- Cumulative Fertility at various
Attained Ages. North, Nigeria, 1990.

94A

The use of total fertility rate as a measure of cumulative fertility assumes a
constant age specific fertility rates. However, fertility decision is dynamic not static.
women may revise their fertility decisions based on both economic, health and biological
circumstances realizing their long term fertility goals. The usual practice of measuring
total fertility eight and four years before a survey and comparing the two fertility rates
may be an inaccurate measure of either an increase or decrease in fertility. We need to
follow the younger women into the future to examine how their fertility patterns are
changing with time. This calls for a follow up survey to examine fertility at a later date.
Thus the recent claim of an overall fall of about 1.3 in total fertility rate among women in
Nigeria and a decline of about 10 percent in fertility among women in the Southwest and
possibly the Southeast may be an overestimation ( Cohen, 1993; Reinis, Rutstein and

Ajayi, 1991; Oslen, 1994).

Birth Patterns.

Another measure of fertility behavior is the onset and timing of births. The age at
first birth and subsequent birth spacing determines the overall fertility of a woman.
Furthermore, for a sustainable fertility decline we should expect a lower proportion of the
younger women to have their first births at each of the attained ages. Table 5.2 shows the
percentage of women who have had their first births at various attained ages.

As Table 5.2 demonstrates, for the all women sample, the percentage of women
with their first births at age 18 rises uniformly from the oldest women to those ages 28-

32 before declining over the next two younger women. We make similar observations for

95

first births at age 23 and 28. However, first births at age 33 rises steadily from the oldest
to the youngest women.

The lower panels of Table 5.2 present the distribution of first births for the
regional samples. In the Southeast, first births at ages 18 initially rises from women ages
43-47 to women in the 38-42 age group but thereafter declines steadily from oldest to the
youngest women. On the other hand , first births at age 23 does not follow any definite
pattern, it fluctuifes a lot At higher ages 28 and 33 however, the percentages are fairly
stable for women ages 28-42 and 3342 respectively. We also observe a postponement
of birth births at ages 18 but a catch-up at ages 28 and above. However, the five-year
age span between 18-23 years may be too long to detect shifts in the median agorat first
birth. ' 1 -;

lln the Southwest, the percentage of women with first births at age 18 flrrrttuates
initially. between women ages 33-47, but thereafter it-declines steadily from thesrldest to
the yéungest women. We make similar observations for first births at age 23. :Ihat at
age 28.also rises. eadily but declines with women ages 28-323'However, first births at
age 33, on the other hand rises uniformly from the oldest to the youngest women. We
also observe a postponement of first births at age 18 but a catch-up at age 28 and above.

In the North we have a different picture about the percentage distribution of
women with first births at various attained ages. The percent of women with first births
at all ages rises uniformly from the oldest to the youngest women. In comparing the

three regions we observe that at younger ages a higher percentage of women innthe North

3.x--- . j ‘_ .-,_ ‘- _ _

96

TABLE 5.2
Percentage of Women with First Births at Various Attained Ages
By Five-Year Age Group. Nigeria, 1990.

 

 

 

By Age

18 23 28 33 N
W
Age Group
18-22 30.4 - - - 1831
23-27 33.7 72.6 - - 1684
28-32 39.1 77.5 91.4 - 1587
33-37 35.2 75.4 92.3 96.4 955
38-42 32.1 67.8 85.5 93.0 951
43-47 29.9 65.9 83.0 92.3 572
SOUTHEAST
Age Group
18-22 22.0 - - -- 487
23-27 27.2 67.9 - -- 427
28-32 - 37.5 75.3 91.1 - 392
33-37 38.2 73.1 91.3 96.0 275
38-42 41.8 74.1 92.4 96.8 251
43-47 35.5 74.0 87.6 96.4 169
SOUTHWEST
Age Group
18-22 10.2 - - - 609
23-27 15.4 57.4 - - 519
28-32 25.6 71.0 89.9 - 445
33-37 25.9 78.6 97.2 100.0 290
38-42 19.9 69.4 91.8 97.6 291
43-47 _ 20.0 72.9 90.0 96.5 170
m
Age Group
18-22 52.0 - -- - 755
23-27 50.4 86.0 - -- 738
28-32 47.9 82.5 92.5 - 750
33-37 40.0 74.6 89.2 94.1 ' 390
38-42 34.7 62.8 76.8 87.3 409
43-47 33.0 54.9 74.7 86.3 233

 

Source: Calculated from Demographic and Health Survey, Nigeria, 1 990

97

have had their first births. However, at older ages the younger women in the North

compared well with those in either the Southeast or the Southwest.

Marital Patterns

Marital status is closely related to fertility, and changes in marital status track the
changes in fertility, especially first births. In addition the length of time a women is in
union indicates the degree of her exposure to the risk of pregnancy. If fertility has indeed
fallen, we should expect that a higher percentage of older women would first marry at
each of the attained ages than the younger women. In Table 5.3 we document the
percentage of women who have ever been married at various attained ages for all women
ages 18-47 and also for the three regions.

In the overall, women ages 18-47, the percentage of women who first marry by
age 18 rises steadily from the oldest women and peaks with women ages 28-32 and
declines over the next two youngest women. We make similar observations for first
marriages at age 23. However, first marriages at ages 28 and 33 remain almost constant
among all women since there is almost universal marriage at age 28. Overall we see a
catch-up in marriages between ages 18 and 28. Women are postponing marriages at age
18 but by age 28 and above we witness almost a universal marriage among the different
age groups.

In the lower panels of Table 5.3 we look at the percentage of women who first
marry at various attained ages by region of residence. First we observe that for first

marriages at ages 18 and 23 the percentage of women who first marry in the Southeast

98

and the Southwest fluctuates among the first three age groups of oldest women and
thereafter declines steadily hour the oldest to the youngest women. First marriages at age
28 in the Southeast follows no definite pattern but that in the Southwest exhibits an
inverted U-shaped distribution. First marriages at age 33 in the Southeast declines
steadily from the oldest to the youngest women, but in the Southwest remains almost
constant around 100 percent. We have fewer women who first marry at age 18 in the
Southwest than the Southeast, however, at ages 23 and over women in the Southeast
sometimes report fewer marriages than their counterparts in the Southwest. We also
observe that the younger women in the Southeast and the Southwest are postponing
marriages at age 18 and 23 but there is almost universal maniage at ages 28 and 33.

In contrast, for women residing in the North, we observe a steady increases in the
percentage of women who first marry at ages 18 and 23 from the oldest women to the
women ages 28-32 and declines slightly over the next relevant younger women. Even
though the percentage distribution eventually declines, we still observe that a higher
percentage of the younger women first marry at ages 18 and 23 than the oldest women.
At ages 28 and 33 we observe a steady increases in the percentage of women who first
marry from the oldest to the youngest women. In the North we see women postponing.
first marriages at only age 18 but we see a catch-up at ages 23 and beyond.

In comparing the three regions, we observe that on average for first marriages that
occurred at ages 18 and 23 women residing in the Southwest report the lower percentages
than women in the Southeast. Women in the North report the highest percentages of

married women at ages 18 and 23. However, for first marriages at ages 28 and 33 the

oldest women ages 43-47 in the North report slightly lower percentages of first marriages
than women in the Southwest and the Southeast. There is almost a universal marriage at
ages 28 and 33 so the percentages in the three regions are very similar and closer to each
other. Notice in particular that in the Southwest and the Southeast the youngest women
eventually report lower percentages of marriage than the oldest women, but this is not the
case in the North.

Women may postpone their first births for a variety of reasons. They may not be
financially ready for the birth of their first child or job opportunities may conflict with the
birth of a child. They may also postpone their first birth to complete their schooling.
Work in the formal sector may require a new woman employee to work for at least two
years before maternity leave with benefits could be paid. A newly employed woman may
thus find it advantageous to postpone her first for financial security reasons. Thus a
women in a modern competitive society may postpone first birth for the above reason.
This may be the underlying factor for younger women in the Southeast and the Southwest
postponing their first births. However, because of the relatively underdevelopment in the
North, the opportunity cost of child birth may not be higher and thus the incentive to

postpone first birth after marriage may not be as great as in the two southern regions.

100

By Five-Year Age Group. Nigeria, 1990.

TABLE 5.3
Percentage of Women Ever-Married at Various Attained Ages

 

 

 

By Age
18 23 28 33 N
ALL WOMEN
Age Group
18-22 51.0 -- -- -- 1831
23-27 57.8 83.4 - - 1684
28-32 67.9 90.5 97.7 -- 1587
33-37 64.6 89.8 97.7 98.7 955
38-42 63.5 89.1 97.8 99.5 951
43-47 60.1 88.3 97.2 99.5 572
SOUTHEAST
Age Group
1 8-22 3 7.4 -- -- - 487
23-27 47.3 74.9 -- -- 427
28-32 57.7 85.7 94.4 - 392
33-37 59.3 83.3 93.8 95.6 275
38-42 62.2 86.9 96.4 98.4 251
43-47 59.8 85.8 95.9 99.4 169
SOUTHWEST
Age Group
18-22 22.5 - - -- 609
23-27 26.4 70.3 - - 517
28-32 40.2 82.5 97.5 - 445
33-37 47.2 87.9 99.0 100.0 290
38-42 37.5 83.2 96.6 99.7 291
43-47 42.4 85.9 98.2 100.0 170
NORTH
Age Group
18-22 83.5 -- -- - 735
23-27 85.9 97.4 -- -- 738
28-32 87.9 97.9 99.5 - 750
33-37 81.3 95.9 99.5 100.0 390
38-42 82.9 94.6 99.5 100.0 469
43-47 73.4 91.8 97.4 99.1 233

 

Source: Calculated fiom Demographic and Health Survey, Nigeria, 1 990

101

Conclusion

In the above discussion we have established the fact that education for women has
increased for the younger women. If fertility has indeed fallen we should see that the
younger women have fewer mean CEB at each of the attained age, the younger women
should first marry and have their first births at much later age than the older women.

However, in looking at cumulative fertility at various attained ages, we do not
observe an overall fall in fertility. Fertility eventually declines with the last two
youngest cohorts, but even then at ages 28 and over, the cumulative fertility of the
youngest cohorts is sometimes higher than that of the oldest cohorts.

In the Southeast we see a decline in cumulative fertility at ages 18 and 23,
however, for cumulative fertility at ages 28 and beyond younger women seem to report
higher mean CEB than the older women. In sharp contrast in the North, we observe a
rising trend in cumulative fertility at various attained ages among women.

An investigation of first births and first marriages at various attained ages also
revealed similar results. We do not see a declining proportion of younger women
marrying and having their first births. At younger ages 18 and 23 we observe a slight
decline in the percentage of young women who marry and have their first births but there
is a late catch-up at ages 28 and beyond. Whilst this is true in the overall and the

Southeast and Southwest, in the North we see a catch-up at age 23.

102

5.4. FAMILY PLANNING POLICIES AND CONTRACEPTIVE USE
Family Planning Policies

Family planning policies and the encouragement of women to use contraceptives
for either birth spacing or stoppage has been the main direct influence on fertility. Since
1983, organized family planning services have received great governmental support.
Family planning policies have now become a part of the state public health system. In
addition family planning policies in Nigeria have received tremendous support from its
Federal Ministry of Health, USAID, the World Bank and UNFPA (Feyisetan and

Ainsworth, 1996).

Organization And Availability of Family Planning Services

Nigeria has a federal system of government with three tiers, Federal, state and
local government. Policies and guidelines concerning family planning services are set by
the Federal government. On the other hand, policies for the provision of services are
determined at the state level and the actual implementation and provision of family
planning services are handled by the local government areas (LCAs).

The organization of family planning facilities is structured on a three tier system.
At the local level is the primary tier charged with among other things the provision of
maternal and child health (MCH) care. At the state level we have the secondary tier
which is composed largely of district hospitals charged with the provision of curative
services. The federal level is the tertiary tier, which comprises of all the teaching

hospitals.

103

According to the NDHS, 1990 data about 9.8 percent of currently married women
in the rural areas live in communities served by a health worker who provides family
planning services. Mobile clinics and market outlets together provide 4.5 percent of the
family planning outreach services. Health workers provide family planning services to
13.3 percent of the women in the Southeast, and about 8.7 percent of women in the
Southwest. In the North we have about 15.3 percent of women in the Northwest and 1.1
percent in the Northeast who receive family planning services from health care
providers. (Table 10.1, NDHS, 1992)

About 31.4 percent of all currently married rural women live within 5 miles of a
facility that provide family planning services. As many as 41.6 percent of women in the
Southeast live in communities where informants were unable to identify a stationary
facility which provides family planning services. On the other hand in the Southwest, all
women live in communities with a stationary facility providing farmly planning services
(see Table 10.1, NDHS, 1992 page 119). Also according to the NDHS, 1990 data about
37 percent of users of family planning supply methods live within one mile of a facility
- oflering family planning methods compared to 9 percent nonusers (see also F eyisetan and

Ainsworth, 1996).

Contraceptive Use
In the NDHS, 1990 data information about the knowledge and use of
contraceptive methods was collected by asking the respondent to name ways or methods

by which a couple could delay or avoid pregnancy. If respondent failed to mention a

104

particular method spontaneously, the interviewer described the method and asked if she
recognized it. Both modern and traditional methods were described.

About 45.7 percent of all women have some knowledge of a contraceptive method
and about 34 percent know a source. Among currently married women about 43.6
percent know a contraceptive method and 31.9 percent know a source (Table 4.1, NDHS,
1992). In the urban areas about 70.4 percent of currently married women know a
method, in contrast only 36.3 percent of women in the rural areas know a method . In the
Southeast about 56.6 percent of currently married women know a contraceptive method,
in comparison about 75.6 percent of currently married women in the Southwest know a
contraceptive method. In contrast, in the two northern regions, about 24 percent and
30.3 percent of currently women in the Northeast and Northwest know any method of
contraceptive respectively (Table 4.2, NDHS, 1992)

Among currently married women with no schooling, only about 29.1 percent
know any method of contraceptives. For currently married women with some primary
education about 63.1 percent know a method, and roughly about 67.2 percent of those
who have completed primary know a method. In comparison, about 83.8 and 90.8
percent of currently married women with some secondary and completed secondary or
higher know a contraceptive method respectively (see Table 4.2, NDHS, 1992)

With regard to the use of contraceptives among women, we observe that, about
15.2 percent of all women have ever used a contraceptive method, about 9.0 percent

modernand about 9.3 percent traditional method. On the other hand among currently

105

married women about 14 percent have ever used a contraceptive method, 8.4 percent
modern method and 8.1 percent traditional method. (see Table 4.3 NDHS, 1992)

Among women who are currently using contraceptives about 7.6 percent of all
women are use a method of contraceptives, 3.8 percent modern and 3.8 percent
traditional. Among the currently married women about 6 percent use a method of
contraceptives, 3.5 percent modern and 2.5 percent traditional (see Table 4.4 NDHS,
1992).

In terms of the influence of one’s background characteristics and the use of
contraceptives, we observe that about 14.8 percent of the urban women currently
married use contraceptives, 9.6 percent modern and 5.2 percent traditional methods.
Among those in the rural areas only about 3.6 percent are currently using a contraceptive
method, 1.9 percent modern and 1.7 percent traditional. In the Southeast about 8.8
percent of currently married women use contraceptive, 3 .9 percent modern and 5 percent
traditional. In the Southwest this increases to 15 percent, 10.5 percent modern and 4.5
percent traditional methods. On the other hand in the Northeast and Northwest only 2
percent and 1.2 percent of currently married women use a contraceptive method
respectively. In the Northeast 1.3 percent use modern methods and only 0.7 percent use
traditional methods. In the Northwest 0.7 percent use modern methods and only 0.5
percent use traditional methods.

Among women with no schooling, about 2 percent use any method (1.3 percent
modern and 0.7 traditional methods). On the other hand, for women with some primary

education 7.8 percent use any method (3.9 percent modern and 3.9 percent traditional).

106

Those who have completed primary, 10.5 percent use any method (6.4 percent modern
and 4.1 percent traditional). Among women with some secondary about 17 percent use
any method (9.7 percent modern and 7.3 percent traditional. In comparison, among
women who have completed secondary or higher, about 28.4 percent use modern and
11.7 percent traditional) (see Table 4.5, NDHS, 1992).

Relatively higher percentage of the middle-aged women use contraceptives than
the older women. For example, among all women ages 20-24 and 25-29 about 9.5 and
8.6 percent currently use contraceptives respectively. In comparison, about 7.9 and 4.2
percent of women ages 40-44 and 45-49 currently use contraceptives respectively. We
make similar observations for the currently married women.

There are three main kinds of sources of supply for contraceptive methods in
Nigeria, namely from the government, the private sector and other sources. The
government supplies about 36.7 percent of the modern contraceptives for current users.
The private sector supplies about 47.2 percent of the modern contraceptives for current
users and the rest come from other sources including Mission and Friends, relation and

unknown sources (Table 4.8, NDHS, 1992).

5.5. SUMMARY AND CONCLUSIONS

 

This chapter has examined both female educational and fertility trends and family
planning and contraceptive policies in Nigeria. Our evidence support the fact that female
schooling has increased substantially fi'om the oldest to the youngest cohorts. This is

observed among all women in the survey as well as in the regional and urban samples.

107

We investigated the trends in fertility in general using indicators such as the
number of children ever born, marital and birth patterns and cumulative fertility at
various attained ages for all the women in the present analysis and also for the three
major regions. Using the retrospective birth histories in the NDHS, 1990 data, we
reconstructed cumulative fertility at various attained ages. We observed that at younger
ages 18, 23 and 28, cumulative fertility declines with the youngest women within the
relevant age range, but the decline is not large enough to offset the high fertility of the
oldest women.

In the Southwest and the Southeast, we witness an eventual decline in fertility
among the youngest women at ages 18 and 23. However, this does not indicate a decline
in fertility since at ages 18 and 23 the older women had not completed their childbearing
years and the younger women have just begun their reproductive years. On the other
hand at ages 33, 38 and 43 we do not have the younger women to compare with. We see
no indication of fertility decline in the North. We make similar observations for first
marriages and first births for all the women and in the regions.

Family planning policies and programs have expanded since 1983-1989 when
officially sponsored family planning programs began. However, family planning
programs are still inaccessible to women in the North and in the rural areas. The distance
to the nearest family planning facility is still more than one mile reach to 63 percent users
and 91 percent of nonusers.

Contraceptive use has also gone up slightly among all women and also among

currently married women (from one to six percent in three decades, both modern and

108

traditional methods). Nearly 46 percent of all women have some knowledge of a
contraceptive method, this compares well with 43.6 percent among currently married
women. Among women with no schooling only 29.1 percent know a method and nearly
90.8 percent of those who have completed secondary or higher know a method.

In sum one may conclude that the recent claim of a fertility decline in Nigeria of
the magnitude of 1.3 in the total fertility rate between eight and four years before the
survey may be an overestimation of the actual decline in fertility. There has been some
expansion in the family planning policies and programs, however, services must be made
more accessible to both users and nonusers. More educated women have better access to
the information and use of contraceptives.

For a significant reduction in fertility levels, however, education must be more
accessible to women with regard to the knowledge and use of contraceptives. To
investigate how much of the decline in fertility if any, is accounted for by the increases in

women-’5 education we do a multivariate regression analysis in our next chapter.

109

CHAPTER 6
THE DYNAMIC RELATIONSHIP BETWEEN FEMALE SCHOOLING AND
FERTILITY
6.1. INTRODUCTION.

Fertility transition has been slow and limited in sub-Saharan Africa. This is in
part due to lack of development and unique cultural and religious practices that promote
the demand for bigger families (Caldwell and Caldwell, 1987, 1990; World Bank 1984,
1986). In spite of these difficulties, some studies have found differentials in recent
fertility or total fertility and desired family size among different socio-economic groups
in sub-Saharan Africa ( Cochrane and Farid, 1990).

Using Nigeria Demographic and Health Survey (NDHS) 1990, this chapter
examines the evidence of a fertility decline in Nigeria. At the time of the survey, official
family planning had been in force in Nigeria for only one year. In addition only about six
percent of currently married women use contraceptive method (3.5 modern and 2.5
traditional). The NDHS, 1990 data also suggests that female schooling has increased
dramatically for the younger women in the survey. For example, among women ages
18-47, women who are ages 43-47 have only 1.47 mean years of completed schooling,
this compares with 5.54 means years of completed schooling for women ages 18-22 (see
chapter 4). In the light of this tremendous increases in female schooling in recent years,
one would expect fertility to have started to decline. In chapter 4 we found rather limited
evidence of this. Fertility patterns are likely to differ by female schooling, the subject of

this chapter. There has not been any systematic quantification of the influence of female

110

schooling on fertility in Nigeria. As mentioned in our literature review chapter, the use
of the number of children ever born as a dependent variable in regressions that seek to
examine the relationship between female schooling and fertility has censorship problems,
since younger women in the survey have not had the same length of exposure to the risk
of pregnancy and childbearing as the older women in the survey. This chapter provides
an alternative measure of cumulative fertility. Using the birth histories of women in the
survey, the cumulative fertility for each woman at age 18, 23, 28, 33, 38 and 43 was
reconstructed. These variables will serve as our dependent variables.

We have reason to believe that the increasing trend in female schooling for
younger women will alter both the demand and supply of children among the younger
women and among the better educated This is because the negative correlation between
female schooling and fertility is well established in the empirical work on fertility
(Cochrane, 1979; Ainsworth, Beegle and Nyamete, 1996).

This chapter examines in some detail, using both ordinary least squares and logit
regression techniques, about how much of the decline in fertility in Nigeria is accounted
for by female schooling alone. Later we introduce dummies for the different cohorts to
investigate the cohort effect on fertility. We also introduce dummies for community
variables such as urban residence, religious affiliation and type of marriage to
investigate the influence of such variables on fertility decision making.

The rest of the chapter is organized as follows: Section 11 presents the theoretical
and empirical model of the present study. Section III also presents the methodology of

the study. Section IV discusses the results of both the multivariate least squares

111

regression of the determinants of fertility and the logit regressions on first births and first
marriages at various attained ages. Beginning with the children ever born, this section
first looks at the relationship between the number of children ever born and single years
of female schooling. We fiuther investigate the relationship between cumulative fertility
at various attained ages and female schooling, birth cohort and community variables.
The section also looks at the relationship between female schooling and age at first birth
and age at first marriage. Finally the section also investigates the relationship between
both female and husband's education on fertility for the ever-married women. Section V

concludes the chapter.

6.2.1. THE THEORETICAL MODEL

 

Following the Quality-quantity economic model of household production (Becker,
1981; Becker and Tomes, 1973; Willis, 1973) as applied to contraceptive use and
fertility (Rosenzweig, 1990 ; Rosenzweig and Schultz, 1993, 1985: Schultz, 1997), we
assume that households choose to allocate resources in order to maximize their lifetime
utility which depends on the consumption of both market and non-market goods. Non-
market goods comprise of the number of their children C, the quality of their children Q
(quality as measured by the average education and health of their children), the average
leisure activities of both parents, L, and L,” and all market commodities represented by
a composite commodity G: The household lifetime utility function has the usual

neoclassical properties and is given by:

112

U = U(C9 Q9 L119 Lwr G) (l)

The household is assumed to be both a consumption and a production unit. It is
assumed that each argument of the utility function is produced with both market goods X,
and non-market time of both parents. Thus the production for each of the argument is

given by;

Ai = A1 (Xi, thr t-w” 11) (2)

where i = C, Q, Lh, Lw, G. and p. is a couple specific traits which influence production
possibilities. In the case of the production of children, 11 might be exogenous genetic or
environmental factors that affect a couple's production of children, which is called
"fecundity". (Rosenzweig and Schultz 1983, 1985, 1987).

The household maximizes its lifetime utility function subject to both a time
budget constraint and an income budget constraint. We assume for the moment that the
time of the husband and that of the wife are mutually exclusive. Also both couple
allocate time for both market and non-market activities. The time constraint for each

couple is given by:

T.

J

=5... 1' 23191 + 21 L1 (3)

113

wherej = h, w, i = C, Q, Lh. Lw, G, and t,“ and t,- arc market and non-market time. The
income budget constraint is also given by::

Y = thmwh 1’ l-wmww +V (4)

where Wj is the lifetime wage rate received by each member of the family so that
lifetime market income is equal to the wage rate times market labor supply. V is non-
labor income. of both household members. The full-income budget constraint is thus

given by

F =T,,wh + r,,ww +v (5)

If pc represents the price of each child inputs, and Wh and Ww is the opportunity
cost of leisure for the husband and wife respectively, and p5 is the price for the
composite market commodity then the full-budget constraint facing the lifetime utility

maximizing household is given by :

pc(CQ) + Wth + W,,,Lw + pGG = ThWh + TWWw + V (6)
This gives rise to a nonlinear budget constraint. However, the assumption of a more
convex utility function ensures an interior solution to the constrained utility maximization
problem.

The reduced form demand equations for the household produced commodities are

therefore the solution to the maximization of equation (1) subject to the full-budget

114

constraint given by equation (6) The reduced form derived demand functions for each of

the argument in the utility function is thus given by:
Zi = Z 1 (Pt, wh9 wwr V9 62 g )9 (7)

where P" is a vector of average prices of the market good and public services available to
the household including that of the price of child inputs Child inputs may include the
time of both couple, place or region of residence among other things. Wh, WW, and V
are as defined above and G) is household specific traits that are a subset of 11, and g is

an unobservable random disturbance term.

6.2.2 EMPIRICAL SPECIFICATION OF THE MODEL

 

Children like any other household production good impose both time and income
constraints on the household. On the other hand, children may be valued as productive
assets for their labor while young and as investments that may yield returns to parents in
the future. The desire to lower fertility below its natural level depends on both the direct
cost of raising children as well as the imputed value of time cost of both parents. Given
the traditional assumption that women allocate more time to childbearing than men do, it
is therefore, likely that the desired fertility will be lower among the higher wage and thus
the better educated (see chapter 3). Maternal education is also likely to play other roles
such as improved information on the efficient use of contraceptives. Given that the

demand for children is a joint decision of the household, both husband's and wife's

115

education, and hence income may influence the desired demand for children. However,
the substitution efi’ect of the woman's income will outweigh the income effect of the
husband's earning. Measures of both couple's lifetime wage earnings are given by their
level of education. Couple specific traits influencing the number of children may include
the relative age of the woman, community variables including type of union, area of
residence and religious affiliation.

Thus the derived reduced form demand for children may be reformulated as:

2* = z*( 13,, Eh, BCD, U, R, M, 6,6),

where EW, 13,, is the education of wife and husband respectively, BCD is birth cohort
dummies, U isarea of residence, R is religious affiliation, M is type of union, 6 is a
couple specific traits which is a subset of 9 and e is an unobservable random
disturbance term.

The reduced -form model of fertility determinants we estimate using both OLS

and logit is in general of the form:

3’11 = 0‘0 ‘1' (1le +111]

yij = Bo '1' BrEw 1' 13ch1) TCij

yij 1.0 + 2.le + lzBCD + 2.3Urban + MReligion + iii

y” = 80 + 6,EW + 52E.| + 83BCD + 84Urban + 85Religion + 86Polygamy +8“

116

where :

(19, B0, A-09 and 50

Ew

BCD

Urban

Religion

Polygamy

111,3, C119 Enj- and 811,

represents the dependent variables, cumulative fertility at

various attained ages, first births and first marriages at various
attained ages where appropriate, i indexing the dependent variable

and j indexing the number of observations.

are intercepts

is woman's education in single years dummies up to ten years and
eleven or more years of schooling.

represents husband’s education in its various specification
represents birth cohort dummies of the woman representing some
'of the household specific traits.

is whether the woman lives in an urban area also capturing some
of the environmental influence on fertility

is the woman's religion also capturing community influence on
fertility

represents polygamous union which is measured by number of

other wives.

are error terms capturing unexplained factors affecting fertility and

also omitted variables from our regression.

117

Omitted Variables and Measurement Error Problems.

The nature of the data we have does not permit us to include all the variables that
may affect fertility that theory suggests. The two main types of data problems, are that
which may originate from omitted variables and that which may be due to measurement
errors.

Omitted Variables

In our first sets of regressions we use only women's education as a measure of
potential lifetime wage and variables that measure couple specific traits. We have
omitted non-labor income because our data does not have observations on non-labor
income. Non-labor income may help a couple to probably acquire more education for
themselves or more investments in their children. Thus non-labor income may be
correlated with the education of the wife. Therefore, the coefficient on women's
education may be biased, the extent of the bias depending on the direction of correlation
between the omitted non-labor income and education.

To solve this problem, in our second set of regressions however, we controlled
for husband's in addition to that of the wife's when looking at the ever-married women.
We had to restrict our analysis to the ever-married women for missing values problems.
Even though husband's education may not be a perfect proxy for non-labor income, it

captures some of the income effects from non-labor income if we had data.

118

Measurement Errors.

The other major problem is that of measurement errors. Chapter 4 of this study
discussed in great details some of the potential measurements errors in the NDHS, 1990.
These may be both sampling and non-sampling errors. Whilst sampling errors can be
dealt with easily by consistency and reliability tests on the data, that of non-sampling are
not easy to detect.

The major measurement errors we encountered were age heaping and
underreporting of births by the oldest women in the survey. Since errors in the reporting
of births affect cumulative fertility at various attained ages, which is our dependent
variable, the errors can be absorbed into a composite error term on the right hand side.
There will not be any estimation problems if the composite error term is not correlated
with the regressors. However, from chapter 4 we observed that the better educated
women report with greater accuracy than the uneducated. In addition the older women
had problems with memory lapse, so the error term is likely to be correlated with both
age and education. However, since age of respondents and education appear as a right-
hand side variables, this may lead to bias and inconsistent coefficient estimates. To avoid
this problem we eliminated the oldest women ages 48-49 in all of our regressions and

also women ages 43-49 in some of our regressions when appropriate.

119

6.3. METHODOLOGY

 

6.3.1 Measurement of Variables

Our primary measures deal with education, fertility, age at first marriage and age

at first birth.

Dependent Variables

The measure of fertility we use is cumulative fertility at various attained ages.
Using the retrospective birth histories of the respondents, we reconstructed the
cumulative fertility at various attained ages: 18, 23, 28, 33, 38 and 43. These will serve
as our dependent variables, replacing the traditional use of number of children ever born
as a dependent variable. The NDHS, 1990 data included age at first marriage, from this
we reconstructed a dummy variable equal to one for women who had married by various
attained ages likewise for first births. These are the dependent variables for the

examination of schooling on age at first birth and age at first marriage.

Independent Variables
Women '3 Education

The NDHS, 1990 data included completed female education in single years. To
examine in more details the relationship between female schooling and fertility, and to
capture the nonlinear relationship between fertility and education, we created dummies
for single years of schooling up to ten years and then a dummy for eleven or more years

of schooling.

120

Cohort Eflects

The desire of women to have children may vary by cohort. Women in relatively
large cohort group may find it advantageous to delay the onset of childbearing and
acquire more education to enter the labor market earlier, the reverse is true for women in
a relatively small size cohort (Easterlin, 1968). Younger cohorts may benefit more from
improved contraceptive technology, increased educational quality and changes in social
acceptance of women in the formal labor force, making it easier to have lower fertility in
the long- run. Thus by adding the dummies for the cohorts into our regression analysis,

we hope to capture the role of cohort-specific factors in fertility decision making.

Area of Residence

To capture the influence of area of residence on fertility we include urban dummy
in the regressions. The NDHS, 1990 data, also includes a variable on area of residence,
urban and rural. It is generally believed in the literature that because of relatively higher
level of living and formal labor market opportunities urban living offers, fertility tends to
be lower among women residing in the urban relative to those living in the rural areas

(Ainsworth, 1989; 'Singh and Casterline, 1985).

Religion
It is also believed in the literature that a woman’s religious affiliation may alter
her demand for children. Studies have found higher fertility rates for Muslims than non-

Muslims (Yankey 1961; Rizk, 1963, 1973; Sinha, 1957; Caldwell, 1968). In the West,

121

studies have shown that Catholics have higher fertility than non-Catholics (Glass, 1968;
Chou and Brown, 1968; Ryder and Westoff, 1971). In Nigeria, empirical work have also
found Christians to have lower fertility than non-Christians. Muslim and traditional
worshipers to be more fertile than Christians (Obadike, 1968; Orubuloye, 1981). The
NDHS, 1990 data also includes a variable for religious affiliation of the respondents. We
include dummies for women who are Protestants, Catholics and Muslim with those with

traditional and no religion as our excluded variables.

Type of Marriage

Monogamy and polygamy exists side by side in many African countries
including Nigeria Islam and African Traditional religion promote polygamous marriage
so it a common practice in many Afiican countries. It is believed in the literature that
individual women in polygamous union have lower fertility, since they have to share
their husbands with other women. and since a large family size can be achieved by fewer
children per woman (Namfua, 1981; Orubuloye, 1981). The NDHS, 1990 data contains
a variable for type of marriage. Respondents were asked if they were in a polygamous
union and if so the number of other wives the respondent’s partner currently had. The
base was women who were currently married. We created a dummy for respondents
who reported to have one through nine other wives into polygamy and use this as an
additional regressor. Polygamy at the right hand side poses a potential endogeneity

problem. This is because in most agricultural societies where polygamy is practiced,

122

polygamy has been found to be positively related to the productivity of wives on the

farm (Jacoby, 1995).

Husband's Education

The NDHS, 1990 data, has no measure of household income or consumption or
household expenditure. However, it includes current husband's education in completed
years, so we decided to use husband's education as a proxy for household income. If
women bear the greater part in childbearing and childraising, then the impact of maternal
education is likely to be larger than that of the husbands education. On the other hand if
even the husband spends no time raising children, as long as both husband's and wife's
leisure are complements within the household, the husband's education will also have a
negative impact on fertility. To examine the impact of husband's education on fertility
we have dummies representing husbands education for 1-5 years, six years, 7-10 years
and eleven or more years of schooling. As a result of the fact that some of the women in
the survey are not married, and some women could not recall their husbands education
with accuracy We have missing variable problems with regard to husband education.
' Consequently, husband’s education is used only in regressions for the ever-married
women. The coefficient on husband's education may be biased downward because of
random misreporting or displacement of husband's education by respondents or because

current husband may not be a relevant factor for earlier fertility.

123

6.3.2 Methodology

Methods used in this present analysis are multivariate least squares and logit
regression. Even though our dependent variable, cumulative fertility at various attained
ages, takes on only positive values, and more efficient estimation procedures like Tobit or
Poisson Count regression could be used, we decided to use the easy to interpret ordinary
least squares regression analysis (see Ainsworth, 1989). However, because the first births
and first marriages at various attained ages are binary, we decided to use logit

regressions.

6.4. EMPIRICAL RESULTS.

 

We present both Ordinary Least Squares (OLS) and Logit regressions for various
specifications of fertility equations. As dependent variables we use cumulative fertility
at ages 18, 23, 28, 33 and 38 and binary variables indicating whether first births and first
marriages have taken place by age 18 or 23. As a result of almost universal first births
and first marriages at ages 28 and over, the Logit regressions are presented for first births
and first marriages 'at ages 18 and 23. At the right hand side our explanatory variables
include female schooling in single years, dummies for a woman’s relative cohort, urban
residence and religious affiliation, women in polygamous marriage, with our omitted
variables being women with no schooling, rural residence, women in the 43-47 age
cohort, traditional and no religion and women in monogamous marriage. We begin first
by looking at the relationship between children ever born and single years of female

schooling. The regressions for this chapter is restricted to women in the south, since from

124

our previous chapter it became apparent that fertility has fallen among the younger

women in the Southeast and the Southwest but not in the North.

6.4.1. Education and Fertility: Mean CEB and Coleeted Years of Schoofilg.

We begin by looking at the relationship between female schooling and single
years of female schooling. Figure 6.1 is a graphical representation of mean CEB and
single years of schooling for all women ages 18-47 and also for women in the 18-27, 28-
37 and 38-47 age groups.

Consistent with other empirical work in the literature, we find that the relationship
between the mean CEB is highly nonlinear. Among all women ages 18-47, the mean
CEB initially fluctuates and above that for women with no schooling with 1-3 years of
female schooling. However, with four or more years of schooling we see a continuous
fall in mean CEB with additional years of female schooling. So that from 4.35 mean
CEB with no schooling, the mean CEB falls to 4.11 with four years of schooling, and still
declines firrther to 1.28 with eleven or more years of schooling.

The profile representing mean CEB for completed years of schooling for the
youngest women 18-27 is gentle indicating that the youngest women are the most
educated and have little exposure to the risk of childbearing. We also observe that even
among the youngest women, mean CEB for the first few years of schooling also is
associated with higher fertility. It is only after four years of schooling that we observe a

negative relationship between female schooling. From 2.36 mean CEB for women with

125

no schooling this declines even though not uniformly to 0.57 with eleven or more years
of schooling.

For women in the 28-37 age group, mean CEB starts to decline below that of
women with no schooling only with six or more years of female schooling. Mean CEB
declines from 4.77 for women with no schooling to 2.95 for women with eleven or more
years of schooling. As we get to the oldest women ages 38-47, the impact of female
schooling on mean CEB diminishes. In fact it is only with nine or more years of
schooling that we observe a powerful negative influence of schooling on mean CEB.
Mean CEB declines from 6.22 for women with no schooling to 4.86 for women with
eleven or more years of schooling.

We make several observations between female schooling and mean CEB. First
the first few years of female schooling is associated with rising fertility. It is after a
certain threshold has been reached that female schooling depresses fertility. Among the
youngest cohort this is four years, among the middle cohort this is six years and among
the oldest cohort the threshold is nine years of schooling. The influence of schooling is

greatest among women in the 28-37 age group and mildest among women in the 38-47

age group. I ’-

126

Mean No. of CEB

 

 

 

 

 

 

«h

 

 

 

 

 

 

 

 

01 2345678910114-
museum

 

[-t-raa-a-ra-zr-r-msr-o-ma]

 

Figure 6.1- Mean CEB by Completed years of Schooling

127

6.4.2 REGRESSION ANALYSIS
Impact of Education on Fertility: Female Schooling Only

In order to study the relationship between fertility and female schooling,
cumulative fertility at various attained ages is initially analyzed as a function of women's
schooling in single years. The results are presented in Table 6.1 At age 18, 1-3 years of
female schooling has an unpredictable influence on cumulative fertility, with that of
women with one year of schooling being significantly higher. However, with four or
more years of female schooling, we witness a powerful negative correlation between
maternal education and cumulative fertility. For instance, women with six years of
schooling have 0.24 fewer children and the gap widens to 0.53 fewer children with eleven
or more years of schooling, as compared with women with no schooling.

At age 23, we observe a significant negative relationship between female
schooling and cumulative fertility with six or more years of schooling. On the other hand
at age 28 we observe a powerful negative relationship only with nine or more years of
schooling. At higher attained ages 33 and 38, it is only women with eleven or more years
of schooling who have significantly fewer children than their counterparts with no
schooling.

In comparing the influence of schooling across cohorts we observe that with six
years of schooling, at age 18, women have 0.24 fewer children, but at 23 they have 0.22
fewer children. At higher attained ages however, six years of schooling has no significant
impact on cumulative fertility. On the other hand women with eleven or more years of

schooling, at age 18 have 0.53 fewer children, 1.23 fewer children at age 23, and 1.39,

128

1.27 and 1.50 fewer children at ages 28, 33 and 38 respectively. Whilst the influence of
six years of schooling diminishes at higher attained ages that of eleven or more years of
schooling persists and even stronger at higher attained ages. It is possible that with time
the influence of lower levels of schooling on fertility decisions fades out because of lack
of job opportunities. On the other hand that of higher levels of schooling persists
because it enhances a woman's competitiveness in the formal labor market.

Our results parallels other empirical studies on other Afiican countries (Ainsworth
et a1, 1996; Thomas et a1, 1995; Shapiro et a1, 1997). Many reasons are given in the
literature for the initial rise in fertility with the first few years of schooling. Among
which are better hygienic way of living, good nutritional habits formation, and changing
patterns of traditional fertility behaviors such as breastfeeding. Evidence that link good
nutrition to fertility is however, either limited or not conclusive (Bongaarts, 1980;
Menken et a1, 1981). One plausible reason given is the fact that in most of the African
setting, the quality of the education system is poor and with the high school drop out
rates among girls literacy is not gained with one or two years of schooling. That girls
who drop out after a few years of schooling are those whose parents wanted them to drop
out or just could not keep up. Schooling just like any commodity may well be a choice
variable and that self-selection may be a problem, those girls who drop out after a few
years of schooling are not any different from those who never went to school to begin
with (Ainsworth et a1, 1996; Thomas 1995). We should also take note of the fact that

some of the excluded variables like age, household income from our regressions may in

129

part explain the sometime non relation between the first few years of schooling and

cumulative fertility.

Impact of Education on Fertility: Female Schooling and Age Cohorts

Table 6.2 presents our second set of equations which include female schooling
and age cohorts of the respondents as our regressors. The inclusion of the cohort
dummies does not change the significance of female schooling and has very minimal
influence on the magnitude of the coefficients.

In all the regressions we continue to observe an inverted U-shaped distribution of
the number of children born at various attained ages from the oldest to the youngest
women as we observed in our descriptives in chapter 5. For instance, at age 18, we
observe that relative to the oldest women ages 43-47, cumulative fertility increases
steadily reaches its peak at a significant positive value of 0.13 for women ages 28-32, and
declines over the next two younger age cohorts. Similar observations are made for
cumulative fertility at ages 23 and 28, whilst that at age 33 rises steadily. There is thus
no indication of a fertility decline among the younger women relative to the oldest
wOmen in the survey. We had anticipated however, that with the tremendous increases in
women's education among the younger birth cohorts, this would be accompanied by

declining fertility also among the younger women, but the evidence points to the

contrary.

130

TABLE 6.1 - Impact of Education on Fertility - Female Schooling only

 

 

 

Dependent Variable Cumulative Cumulative Cumulative Cumulative Cumulative
Fertility at Fertility at Fertility at Fertility at Fertility at
Age 18 Age 23 Age 28 Age 33 Age 38
18-47 23.47 28-47 33-47 38-47

Indemndent

Variables:

1 year of Schooling 0.356" 0.615" 0.964" 1.495" 1.634"
(0. 14) (0.29) (0.40) (0.54) (0.76)

2 years of Schooling - 0.106 0.240 0.437‘" 0.362 0.438
(0.08) (0.17) (0.25) (0.34) (0.51)

3 years of Schooling - 0.085 0.041 0.169 0.769" 0.973"
(0.06) (0.13) (0.19) (0.28) (0.41)

4 years of Schooling - 0.109‘" 0.083 0.273 0.529‘” 0.515
().06) (0.14) (0.22) (0.30) (0.47)

5 years of Schooling - 0.272‘ - 0.139 - 0.014 0.222 - 0.160
(0.07) (0.16) (0.26) (0.37) (0.56)

6 years of Schooling - 0.243‘ - 0.217‘ - 0.077 0.094 - 0.027
(0.03) (0.07) (0.1 l) (0.16) (0.23)

7 years of Schooling - 0.235” - 0.177 0.118 0.901 0.998
(0.10) (0.29) (0.44) (0.67) (0.88)

8 years of Schooling - 0.473’ - 0.515"I - 0.325 - 0.549 - 0.446
(0.07) (0.17) (0.32) (0.51) (0.83)

9 years of Schooling - 0.358“ - 0.456‘ - 0.470" - 0.445 - 0.634
(0.05) (0.13) (0.24) (0.37) (0.58)

10 years of - 0.533“ - 0.700‘ - 0.814" - 0.741 - 0.431

Schooling (0.06) (0.18) (0.35) (0.54) (0.94)

11 or more years of - 0.527‘ - 1.229‘ - 1.388" - 1.271‘ - 1.502‘

Schooling (0.03) (0.07) (0. 13) (0.22) (0.36)

Intercept 0.606' 1.927‘ 3.514‘ 4.918“ 6.002‘
(0.02) (0.04) (0.06) (0.08) (0.1 1)

R-squared 0.08 0.10 0.08 0.04 0.04

F( all Covariates) 35.59 33.53 13.48 6.00 3.20

N 4317 3223 2279 1445 880

 

Note: Standard errors in parenthesis. "‘ - p-value < 0.01, " - p-value < 0.05, "‘u p-value < 0.10.

131

TABLE 6.2 - Impact of Education on Fertility - Female Schooling and Cohort Effects.

 

 

 

 

 

Dependent Variable Cumulative Cumulative Cumulative Cumulative Cumulativ-
Fertility at fertility at Fertility at Fertility at Fertility at
Age 18 Age 23 Age 28 Age 33 Age 38
18-47 23-47 28-47 33-47 38-47

Independent Variables:

1 year of Schooling 0.358” 0.618" 0.960" 1.485" 1.637"
(0.14) (0.29) (0.40) (0.54) (0.76)

2 years of Schooling - 0.106 0.227 0.414‘“ 0.345 0.430
(0.08) (0.17) (0.25) (0.34) (0.51)

3 years of Schooling - 0.092 0.028 0.136 0.732” 0.985”
(0.06) (0.13) (0.19) (0.28) (0.41)

4 years of Schooling - 0.110‘“ 0.070 0.222 0.471 0.524
(0.06) (0.14) (0.22) (0.30) (0.47)

5 years of Schooling - 0.276’ - 0.163 - 0.062 0.179 - 0.161
(0.07) (0.16) (0.26) (0.3 7) (0.58)

6 years of Schooling - 0.248‘I - 0.23 8‘ - 0.114 0.067 -0.027
(0.03) (0.07) (0.1-1) (0.16) (0.23)

7 years of Schooling - 0222‘" - 0.187 0.114 0.906 1.009
(0.10) (0.29) (0.44) (0.66) (0.88)

8 years of Schooling - 0.469‘ - 0.544‘ - 0.390 - 0.608 - 0.449
(0.07) (0.18) (0.38) (0.51) (0.83)

9 years of Schooling - 0.359" - 0.477' - 0.525” - 0.497 - 0.627
(0.05) (0.13) (0.24) (0.3 7) (0.58)

10 years of Schooling - 0.526" - 0.721‘ - 0.897" - 0.831 - 0.422

, (0.07) (0.18) (0.35) (0.54) (0.94)

11 or more years of - 0.529“ - 1.257‘ - 1.460“ - 1.343“ - 1.497“

Schooling (0.03) (0.08) (0.13) (0.22) (0.36)

Cohort Effects

18-22 0.034
(0.05)

23-27 ’ 0.097“I 0.068
(0.05) (0.09)

28-32 0.128” 0.212“ 0.318"
(0.05) (0.09) (0.12)

33-37 0.079 0.146 0.508“ 0.420"
(0.05) (0.09) (0.13) (0.15)

38-42 0.048 - 0.073 0.090 0.150 - 0.069
(0.05) (0.10) (0.13) (0.15) (0.17)

Intercept 0.536" 1.853’ 3.275" 4.716‘ 6.043‘
(0.04) (0.08) (0.1 1) (0.13) (0.15)

R-squared 0.09 0.1 l 0.07 0.05 0.04

F( all Covariates) 25.37 25.74 12.16 5.76 2.94

F(cohort) 2.72 3.96 6.91 4.76 0.16

N 4317 3223 2279 1445 880

 

Note: Standard Errors in parenthesis. ‘- p-value < 0.01, "- p-value < 0.05, "up-value < 0.10

Impact of Education on Fertility: Female Schooling , Cohort and Community
Variables

Table 6.3 presents our third set of equations, which in addition to dummies for
single years of female schooling and age cohorts, includes dummies representing
community variables. The community variables include urban residence and religious
affiliation. As Table 6.3 demonstrates, the inclusion of the community variables do not
change the significance of either the educational or age cohort dummies and has very
little on the magnitude of the coefficients.

In all the regressions, women living in the urban areas have significantly fewer
number of children born at various attained ages. Protestant background has an
insignificant influence on cumulative fertility at various attained ages. Catholic influence
is only significantly higher for cumulative fertility at age 23, otherwise it is not
' significant. In comparison, women with Islamic religious background have fewer
children for cumulative fertility at all attained ages. This is, however, contrary to what
we had anticipated. One would have expected that women with Islamic background enter
into marriage early and thus this would provide the incentive for more children given the
longer exposure to the risk of pregnancy and childbearing. On the other hand since Islam
promotes polygamy, it is believed in the literature that women in polygamous marriage
have fewer children. This is because they have to share their husbands with other women
and larger families could still be achieved with each wife having fewer number of
children. Furthermore, since our regression analysis is restricted to women in the South,

women with Islamic background may well be migrants from the North. Since women

133

TABLE 6.3
Impact of Education on Fertility - Female Schooling, Cohort Effects and Community Variables

 

 

 

 

 

Dependent Variable EuTnulative Cumulative Cumulative Cumulative Cumulative
fertility at Age fertility at Age fertility at Age fertility at Age fertility at Age
18 23 28 33 38
18 - 47 23-47 28-47 33-47 38-47

Ind_e2c_ndent Variables:

1 year of Schooling 0.343” 0.567“ 0.887” 1.395“ 1.721 ”
(0.14) (0.29) (0.39) (0.53) (0.74)

2 years of Schooling - 0.124 0.175 0.315 0.178 0.163
(0.08) (0.17) (0.25) (0.34) (0.50)

3 years of Schooling - 0.095 0.004 0.109 0.727“ 0.998“
(0.06) (0.13) (0.20) (0.28) (0.40)

4 years of Schooling - 0.108‘” 0.046 0.215 0.546‘“ 0.764
(0.06) (0.14) (0.22) (0.27) (0.47)

5 years of Schooling - 0. 291‘ - 0.212 - 0.109 0.187 - 0.035
(0.07) (0.16) (0.25) (0.37) (0.57)

6 years of Schooling - 0.224‘ - 0.216‘“ - 0.063 0.253 0.372
(0.03) (0.07) (0.1 1) (0.17) (0.24)

7 years ofSchooling -0.l83‘°‘ -0.149 0.192 1.140'” 1.541‘”
(0.10) (0.30) (0.44) (0.66) (0.87)

8 years of Schooling - 0.426‘ - 0.499” - 0.292 - 0.265 0.171
(0.07) (0.18) (0.32) (0.51) (0.82)

9 years of Schooling - 0.313‘ - 0.412” - 0.391 - 0.224 - 0.115
(0.05) (0.13) (0.24) (0.37) (0.57)

10 years of Schooling - 0.489‘ - 0.683” - 0.774“ - 0.522 0.109
(0.08) (0.18) (0.35) (0.54) (0.92)

11 or more years of - 0.483’ - 1.214‘ - 1.364‘ - 1.064‘ - 0.909“

Schooling (0.07) (0.08) (0.14) (0.24) (0.38)

Cohort Effects

18-22 0.032
(0.05)

23-27 0.095” 0.067
(0.05) (0.09)

28-32 0139' 0.232” 0.353‘
(0.05) (0.09) (0.12)

33-37 0086‘” 0.156 0.526‘ 0.449‘
(0.05) (0.10) (0.13) (0.15)

3842 0.058 - 0.054 0.122 0.204 0.010
(0.049) (0.10) (0.13) (0.15) (0.17)

Community Variables

Urban Residence - 0.088‘ - 0.140“ - 0.265“ - 0.517‘ - 0.996‘
(0.02) (0.06) (0.09) (0.13) (0.09)

Protestant - 0.038 0.081 0.163 0.007 - 0.153
(0.05) (0.10) (0.15) (0.21) (0.29)

Catholic - 0.004 0.196‘” 0.259 - 0.140 - 0.234
(0.05) (0.1 1) (0.16) (0.23) (0.32)

lslarn - 0.126“ - 0.99 - 0.167 - 0.485” - 0.782
(0.05) (0.1 1) (0.17) (0.25) (0.33)

Intercept 0.606‘ ‘ 1.838‘ 3.258‘ 4.515‘ 6.575‘
(0.06) (0.1 1) (0.16) (0.21 ) (0.28)

R-squared 0.09 0.1 1 0.08 0.07 0.09

F( all Covariates) 22.36 21.87 1 1.15 6.68 5.23

F(Cohort and Community) 5.77 5.34 7.08 7.63 9.38

F(Community) 9.55 6.68 7.16 9.27 1 1.76

N 4317 3223 2279 1445 880

 

134

who migrate into the cities are the better educated, self-selection may also be a plausible

reason for the lower fertility among the Islamic women.

6.4.3. PREDICTIONS: The Decline In Fertility Accounted for by Female Schooling

Following Oaxaca’s decomposition (Oaxaca, 1973), we decided to decompose
the decline in cumulative fertility from the oldest to the youngest women into that which
is accounted for by the increases in female schooling and that which is unexplained by
female schooling.

In Table A6. 1 . we predict how much of the decline in fertility from women ages
38-47 to those ages 18-22 that is accounted for by female schooling, controlling for
relative cohort of the women. First using the regression coefficients on schooling for
women ages 18-47 andsecond using coefficient on the regression for women ages 18-22.
We observe that using the coefficients on women ages 18-47, about 25 percent of the
decline in cumulative fertility at age 18 is accounted for by female schooling. When we
use the coefficients on women ages 18-22, we observe that now at age 18, female
schooling accounts for about 23 percent of the decline in fertility from the oldest to the
youngest cohorts.

We make similar predictions for cumulative fertility at age 23. As Table A6.2
demonstrates, first using the regression coefficient on women ages 23-47, we observe that
female schooling accounts for about 46 percent of the decline in fertility from the women

ages 38-47 to women ages 23-27 at age 23. On the other hand when using the

135

regression coeflicient on women ages 23-27, female schooling now makes a much bigger

impact on the fertility decline, accounting for nearly 61 percent of the decline in fertility.
At higher attained ages however, we have fewer cohorts to compare and the

decline in fertility among the older cohorts is small. In this regard, we do not report the

prediction for cumulative fertility at ages 28 and beyond.

6.4.4 Age At First Marriage and Age At First Birth.

 

Age at First Marriage

Table 6.4 presents the regressions for women who had first been married by ages
18 and 23. At age 18, four or more years of maternal schooling is associated with delays
in first marriages relative to women with no schooling. In our second set of regressions,
the addition of the cohort dummies does not change the significance of the coefficients
on schooling and has very little effect on the magnitude. Relative to the oldest women
ages 43-47, women in the 28-32 and 33-37 age groups many early. When community
variables are added, we observe that urban residence delays first maniages at age 18,
whilst none of the religious variables has significant impact on the timing of first
marriages at age 18.

At age 23, the influence of female schooling on first marriages diminishes. Some
primary education has no influence on the timing of first maniages at age 23. However,
it is only with nine or more years of schooling that we observe female schooling delaying
frrst marriages at age 23. We make similar observations with female schooling and the

timing of first marriages with our second and third set of equations. In the second set

136

women ages 28-32 enter into marriages significantly earlier than the oldest women ages
43-47. On the other hand in our third set of regressions neither the cohort dummies nor
community variables has predictive power on the timing of maniage.

The relationship between female schooling and age at first marriage at 18 and 23
is very similar to that of cumulative fertility at ages 18 and 23. This is not surprising

since first marriages and cumulative fertility are closely related.

Age At First Birth

In Table 6.5 we also look at the relationship between first births at ages 18 and 23
and female schooling, relative cohort and community variables. We make similar
observations between female schooling and first births at age 18 as we did with first
marriages at age 18. Among the cohort dummies however, now women ages 23-27, 28-
32 and 38-37 all have their first births significantly earlier than the oldest women ages
43-47, an observation we made in both regressions two and three. Urban residence is the
only community variable which delays the onset of first births at age 18.

At age 23, the influence of female schooling on the timing of first births again
greatly diminishes. In all three regressions it is only women with eleven or more years of
schooling who consistently delay the onset of first births. In our second equations,
among the cohorts, women ages 28-32 and 33-37 have their first births early, whereas in
our third equation only women ages 28-32 have their first births early. Among the

community variables only women with Protestant background have their first births early.

137

We make several observations in this section. First, the first few years of
schooling has no predictive power on either cumulative fertility at various attained ages,
first marriages or first births. It is only after a certain threshold of schooling has been
reached that we observe a powerful negative relation between female schooling and
cumulative fertility and other proximate determinants of fertility.

At younger ages 18 and 23, schooling makes an early impact on fertility decisions
including the timing of age at first marriage and age at first birth. At age 18, it takes
four years of schooling, and at age 23 it takes six years of schooling to make an impact on
cumulative fertility, first marriages and first births. It should also be mentioned that at age
18, most women have not attained higher levels of schooling, with the quality of
schooling in most African countries and late entrance into schooling. It is only at age 23
and above that women are either still going to school or nearing the completion of their
education and thus a greater impact of schooling on fertility. At higher attained ages 33
and 38 the influence of schooling on fertility diminishes greatly. An evidence which is

also consistent with the literature, a possible indication of a late catch-up effect.

138

TABLE 6.4 - Impact of Education on First Marriages at Various Attained ages - Female
Schooling, Cohort Effects and Community Variables (Logit).

 

 

 

 

 

 

 

Dependent Variable First Marriage at Age 18 First Marriage at age 23
Regl Reg2 Reg3 Reg] Reg2 Reg3
Independent Variables: 4 I j
1 year of Schooling 1.035” 1.043” 0.988’” 4.033 4.025 4.041
(0.50) (0.50) (0.50) (4.541) (4.53) (4.53)
2 years ofSchooling 0.150 0.141 0.066 - 0.342 - 0.353 - 0.314
(0.24) (0.24) (0.24) (0.35) (0.35) (0.36)
3 years of Schooling - 0.007 - 0.014 - 0.037 0.508 0.493 0.508
(0.18) (0.18) (0.19) (0.38) (0.38) (0.38)
4 years of Schooling - 0.386” - 0.381“ - 0.386” - 0.273 - 0.273 - 0.260
(0.18) (0.18) (0.18) (0.30) (0.30) (0.31)
5ycars ofSchooling -0.113 -0.114 -0.l79 -0.101 -0.110 -0.091
(0.21) (021) (0.21) (0.37) (0.37) (0.37)
6 years of Schooling - 0.593‘ - 0.591 ‘ - 0.533‘ - 0.278" - 0.277‘” - 0.304"
(0.09) (0.09) (0.09) (0.15) (0.15) (0.15)
7 years of Schooling - 0.699' - 0.623" - 0.505 ‘" - 0.221 - 0.227 - 0.262
(0.28) (0.28) (0.28) (0.62) (0.63) (0.63)
8 years of Schooling - 1.50‘ - 1.463‘ - 1.348’ - 0.521 - 0.514 - 0.559
(0.21) (0.22) (0.32) (0.34) (0.35) (0.35)
9 years of Schooling - 1.128‘ - 1.107‘ - 0.985‘ - 0.771 ‘ - 0.750‘ - 0.817‘
(0.16) (0.16) (0.16) (0.23) (0.24) (0.24)
10 years of Schooling - 1.805‘ - 1.757‘ - 1.668' - 0.943‘ - 0.916‘ - 0.955'
(0.22) (0.23) (0.23) (0.31) (0.31) (0.32)
11 or more years of - 2.261‘ - 2.242‘ - 2.124‘ - 2.081 ‘ - 2.067‘ - 2.121 ‘
Schooling (0.1 l) (0.12) (0.12) (0.12) (0.14) (0.16)
Cohort Effects
18-22 - 0.046 - 0.055
(0.14) (0.14)
23-27 0.168 0.166 - 0.013 - 0.012
(0.14) (0.14) (0.19) (0.19)
28-32 0.263” 0.301“ 0.315‘" 0.295
(0.14) (0.14) (0.19) (0.19)
33-37 0.273” 0.296” 0.205 0.189
(0.14) (0.14) (0.21) (0.21)
38-42 - 0.045 - 0.012 - 0.028 - 0.044
(0.14) (0.14) (0.20) (0.20)
Community Variables
Urban Residence - 0.292‘ - 0.141
(0.08) (0.12)
Protestant - 0.006 - 0.051
(0.14) (0.22)
Catholic 0.166 0.021
(0.15) (0.23)
lslarn - 0.239 0.166
(0.15) (0.25)
Intercept 0.400‘ 0.284‘ 0.392” 2.167‘ 2.061 ‘ 2.009‘
(0.06) (0.12) (0.16) (0.09) (0.170) (0.25)
- 2 log likelihood 5176.98 5150.11 5119.83 2730.47 2721.67 2716.72
Goodness of F it 4316.38 4309.80 4323.27 3199.05 3208.68 3225.90
N 4317 4317 4317 3223 3223 3223
°/emarriedby age 18 41.6 41.6 41.6 81.2 81.2 81.2

 

Note ; Strindard Errors in parenthesis. ‘--p-value < 0.01,' “-- p-value < 0.05, “‘- p-value < 0.10

139

TABLE 6.5 - Impact of Education on First Births at Various Attained ages - Female
Schooling, Cohort Effects and Community Variables (Logit)

 

 

 

 

 

 

Note: Standard Errors in parenthesis. ‘- p-value < 0.01, "--p-value < 0.05, “’-- p-value < 0.10.

140

Dependent Variable First Births at Age 18 First Births at Age 23
Reg] Regz Reg3 Reg] Reg2 Reg3
independent Variables: ’ ‘ 4 R g
1 year of Schooling 0.476 0.484 0.433 0.384 0.382 0.348
(0.40) (0.40) (0.40) (0.55) (0.55) (0.55)
2 years of Schooling - 0.074 - 0.078 - 0.149 0.384 0.368 0.323
(0.24) (0.23) (0.24) (0.32) (0.32) (0.33)
3 years of Schooling - 0.034 - 0.058 - 0.088 0.251 0.239 0.194
(0.18) (0.18) (0.18) (0.24) (0.24) (0.24)
4 years of Schooling - 0.347'” - 0.351‘” - 0.372'” 0.070 0.057 0.016
(0.19) (0.19) (0.20) (0.25) (0.25) (0.24)
5 years of Schooling - 0.687’ - 0.700‘ - 0.763‘ - 0.256 - 0.278 - 0.333
(0.24) (0.24) (0.24) (0.26) (0.26) (0.26)
6 years of Schooling - 0.531‘ - 0.542‘ - 0.510‘ - 0.057 - 0.069 - 0.115
(0.09) (0.10) (0.10) (0.1 1) (0.12) (0.12)
7 years of Schooling - 0.672” - 0.618‘” - 0.537 0.110 0.107 0.064
(0.32) (0.33) (0.33) (0.51) (0.51) (0.51)
8 years of Schooling 1.704‘ - 1.684‘ - 1.608‘ - 0.204 - 0.218 - 0.268
(0.30) (0.30) (0.31) (0.28) (0.29) (0.29)
9 years of Schooling - 1.017‘ - 1.013‘ - 0.927‘ - 0.278 - 0.280 - 0.327
(0.18) (0.19) (0.19) (0.20) (0.21) (0.21)
10 years of Schooling 2.178‘ - 2.147‘ - 2.082‘ - 0.320 - 0.322 - 0.367
(0.14) (0.35) (0.36) (0.28) (0.28) (0.28)
11 or more years of - 2.145‘ - 2.146‘ - 2.054' - 1.579' - 1.589‘ - 1.644‘
Schooling (0.14) (0.14) (0.15) (0.11) (0.12) (0.13)
Cohort Effects
18-22 0.121 0.126
(0.16) (0.15)
23-27 0.325” 0.337“ 0.043 0.050
(0.15) (0.16) (0.15) (0.15)
28-32 0483‘ 0.524‘ 0.265‘" 0.273‘”
(0.15) (0.15) (0.15) (0.15)
33-37 0.355" 0.379” 0.263‘” 0.267‘”
(0.16) (0.16) (0.16) (0.15)
38-42 0.152 0.181 - 0.065 - 0.062
(0.16) (0.16) (0.16) (0.16)
Community Variables
Urban Residence - 0.268‘ - 0.013
' ' ' (0.08) (0.09)
Protestant 0.125 0.318‘”
(0.14) (0.16)
Catholic 0.152 - 0.334‘”
(0.15) (0.17)
lslam - 0.134 - 0.259
' (0.16) (0.18)
Intercept - 0.476' - 0.741‘ - 0.726‘ 1.226‘ 1.120‘ 0.868‘
(0.06) (0.13) (0.17) (0.07) (0.13) (0.18)
-2 log likelihood 4400.28 4382>29 4355.28 3645.35 3634.66 3630.56
Goodness of Fit 4316.76 4299.21 4276.64 3222.98 3226.58 3224.35
N 4317 4317 4317 3223 3223 3223
% with lst births at Age 24.7 24.7 24.7 70.1 70.1 70.1

6.4.5 Impact of Education on Fertility: Female and Husband’s Education with
Cohort, Community Variables (Ever-Married Women).

In this section we begin to look at the influence of both spouses education on
fertility decisions. For each dependent variable we select women who have ever been
married at the relevant age. We have also four specifications, the first equation is the
maternal education only, the second being maternal education and that of their age cohort
position, third we include community variables which include urban residence, religious
affiliation and polygamous union. Finally the fourth set of equations include that of their
husbands’ education in addition to those in the third equation.

Table 6.6 looks at this for cumulative fertility at age 18. Female schooling makes
a significant impact for the ever-married women at age 18 even with two or more years of
schooling. We make similar observations in column two when we control for relative
cohort. Relative to the oldest women, women ages 28-32 and 38-42 have more children
at age 18. We continue to observe an inverted U-shaped distribution of births among the
cohorts. The inclusion of the community variables do not change the magnitude and
level of significance of the female education and cohort variables. Among the
community variables it is only Islam background which has a negative impact on
cumulative fertility.

Women in polygamous union have more children even though the coefficient is
not significant. The inclusion of the husband’s education changes the significance of the
lower levels of maternal education, however, women with five or more years of schooling
still have fewer children. Among the husband’s education it is only women whose

husbands have six years of schooling who have fewer children. This is however,

141

consistent with theory, since women who marry at age 18 are the least educated and thus
may find the opportunity cost of childbearing relatively low. Also it is possible that these
women may marry relatively well-to-do husbands and may have little contribution to
fertility decisions in the household.

In comparing this with the all sample, we observe that the ever-married women
have fewer children than all the women sample at lower levels of schooling but with
eleven or more women in the all sample have fewer children. Among the cohorts too we
observe that the ever-married women have born more children as compared with the all
women sample.

In Table 6.7 we also look at the same set of regressions for cumulative fertility at
age 23. At age 23 we witness that six or more years of female schooling (except seven
years) depresses fertility. We make similar observations when we control for the relative
cohort position of women. None of the cohort variables is significant. Women ages 28-

’ 32 have significantly more children at age 23 than the oldest women. Urban residence
has a significant negative impact on cumulative fertility at age 23 but Catholic influence
rather raises fertility at age 23. Polygamous union has an insignificant negative impact
on fertility. Inclusion of husband’s education affects only the significance of women
with six years of schooling, however women with eight and ten or more years of
schooling still have fewer children at age 23. Among the relative cohort and community
variables we make similar observations as we did without controlling for husband’s
education. It is only women whose husbands have eleven or more years of schooling that

have significantly fewer children at age 23. We again observe that higher levels of

142

schooling has a more negative impact on fertility among all the women than the ever-
married women.

At age 28, as Table 6.8 indicates, the influence of schooling diminishes very
much for the ever-married as we observed for the all women sample. It is only ever-
married women with ten or more years of schooling who have fewer children. At age 28
however, the gap between the influence of maternal education on fertility for the all
women and the ever-married narrows. This is not surprising since at age 28 and above
we witness almost a universal marriage. Women in polygamous marriage have fewer
children but again the result is not significant. None of the husband’s education
dummies has significant influence on fertility.

At ages 33 and 38 as Tables 6.9 and 6.10 indicate, the influence of female
schooling diminishes even further as we noticed in the all women sample too. It is only
women with eleven or more years of schooling who have fewer children. Urban
residence and Islam background continue to make a negative impact on fertility. Women
in polygamous marriage have fewer number of children even though the coefficient is not
significant. Husband’s education has no significant impact on fertility. Among the
cohorts we continue to observe that the ever-married women have more children as
compared with the all women sample.

We had anticipated that women in polygamous union will have fewer children.
However, at age 18 we observe its influence to be positive even though it is insignificant.
At ages 23 and above its influence is negative but again it is not significant. It is possible

that Islam and polygamy are measuring the same effect in our regressions.

143

TABLE 6.6 - Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education‘( Ever-married at Age 18)

 

 

 

 

 

 

Dependent Variable: Cumulative fertility Age 18
Reg 1 Reg 2 Reg 3 Reg 4
Independent Variables:
1 year of Schooling 0.204 0203 0.200 0.221
(020) (0.20) (0.20) (0.20)
2 years of Schooling - 0.236'” - 0224‘” - 0237‘” 0.219‘”
(0.13) (0.13) (0.13) (0.13)
3 years ofSchooling - 0.184‘” - 0.197‘” - 0.203‘” - 0.171
(0.10) (0.11) (0.11) (0.11)
4 years ofSchooling -0.l97‘ -0.l95‘” - 0.193‘” -0.160
(0.11) (0.11) (0.12) (0.12)
5 years of Schooling - 0.403' - 0.401 ‘ - 0.405‘ - 0.368‘
(0.12) (0.12) (0.14) (0.13)
6 years ofSchooling -0217‘ -0.218‘ -0.190‘ -0.155"
(0.06) (0.06) (0.06) (0.07)
7 years of Schooling - 0.117 - 0.098 -0.064 - 0.028
(0. 19) (0.20) (0.20) (0.20)
8 years of Schooling - 0.549‘ - 0.540‘ - 0.506‘ - 0.471‘
(0.17) (0.17) (0.17) (0.17)
9 years of Schooling - 0.282“ - 0.280‘ - 0230“ - 0.194
(0.11) (0.12) (0.12) (0.12)
10 years of Schooling - 0.653‘ - 0.646‘ - 0.603‘ - 0.568‘
(0.18) (0.18) (0.18) (0.19)
1 1 or more years of' - 0.490‘ - 0.488‘ - 0.444‘ - 0.416‘
Schooling (0.09) (0.09) (0.08) (0.10)
Cohort Effects
18-22 0.067 0.066 0.072
(0.09) (0.09) (0.09)
23-27 0.139 0.144 0.159‘“
(0.09) (0.09) (0.09)
28-32 0.180“ 0.195“ 0.199“
(0.08) (0.08) (0.09)
33-37 0.091 0.099 0.099
(0.09) (0.09) (0.09)
38-42 0147‘” 0.153“ 0.148
(0.09) (0.09) (0.09)
Community Variables
Urban Residence - 0.071 - 0.063
(0.05) (0.05)
Protestant - 0.077 - 0.065
(0.08) (0.08)
Catholic - 0.071 - 0.054
(0.09) (0.09)
Islam -0.191” -0.180”
(0.09) (0.09)
Polygamy 0.012 0.012
(0.05) (0.05)
Husband's Education
1-5 years of Schooling - 0.109
(0.08)
6 years of Schooling - 0.103’“
(0.06)
7-10 years of Schooling - 0.106
(0.10)
1 1+ years of schooling - 0.082
(0.08)
Intercept 0.906‘ 0.871 ‘ 0.965‘ 0.988‘
(0.03) (0.07) (0.10) (0.10)
R-squared 0.04 0.04 0.04 0.05
F( all Covariates) 5.99 4.51 3.93 3.45
N 1798 1798 1 798 l 798

 

144

TABLE 6.7- Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education ( Ever-married at Age 23)

 

 

 

 

 

 

Dependent Vanab' 1e: Cumulative fertility At Age 23
Reg 1 Reg 2 Reg 3 Reg 4
Independent Variables: r T
1 year of Schooling 0.418 0.423 0.367 0.383
(0.29) (0.29) (0.29) (0.29)
2 years of Schooling 0.328‘” 0.317‘” 0.244 0.232
(0.18) (0.18) (0.18) (0.18)
3 years of Schooling - 0.029 - 0.038 - 0.070 - 0.078
(0.14) (0.14) (0.14) (0.14)
4 years of Schooling 0.048 0.039 0.018 0.015
(0.15) (0.15) (0.15) (0.15)
5 years of Schooling - 0.124 - 0.144 - 0.200 - 0.212
(0.18) (0.17) (0.17) (0.17)
6 years of Schooling - 0.171“ - 0.186“ - 0.154" - 0.128
(0.07) (0.07) (0.08) (0.08)
7 years of Schooling - 0.124 - 0.134 - 0.072 - 0.020
(0.31) (0.31) (0.31) (0.31)
8 years of Schooling - 0.475“ - 0.497” - 0.419“ - 0.389”
(0.19) (0.19) (0.19) (0.20)
9 years of Schooling - 0.298” - 0.314“ - 0.212 - 0.155
(0.14) (0.14) (0.14) (0.15)
10 years of Schooling - 0.575‘ - 0.587‘ - 0.524‘ - 0.422“
(0.20) (0.20) (0.20) (0.21)
1 1 or more years of - 0.864‘ - 0.890‘ - 0.820‘ - 0.676‘
Schooling (0.09) (0.09) (0.10) (0.12)
Cohort Effects
18-22
23-27 0.037 0.039 0.035
(0.10) (0.10) (0.10)
28-32 0.160 0.199” 0.181'"
(0.10) (0.10) (0.10)
33-37 0.105 0.121 0.144
(0.10) (0.10) (0.14)
38-42 - 0.089 - 0.065 - 0.071
(0.10) (0.10) (0.10)
Communig Variables
Urban Residence - 0.205‘ - 0.198‘
(0.06) (0.06)
Protestant 0.093 0.093
(0.11) (0.1 1)
Catholic 0.236“ 0.232”
(0.11) (0.1 1)
lslarn - 0.130 - 0.127
(0.12) (0.12)
Polygamy - 0.055 - 0.064
(0.06) (0.06)
Husband's Education
1-5 years of Schooling 0.086
(0.10)
6 years of Schooling 0.049
(0.08)
7-10 years of Schooling 0.101
(0.13)
11+ years of schooling - 0.199“
(0.10)
Intercept 2.124‘ 2.077‘ 2.093‘ 2.088‘
(0.04) (0.08) (0.12) (0.12)
R-squared 0.04 0.05 0.05 0.07
F( all Covariates) 10.04 8.64 8.71 7.76
N 2619 2619 2619 2619

 

145'

TABLE 6.8 - Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husband's Education ( Ever-married at Age 28)

 

 

 

 

 

Dependent Variable: Cumulative fertility At Age 28
Reg 1 Reg 2 Reg 3 Reg 4
1nmdent Variables: ’ L "
l yem of Schooling 0.891 “ 0.888“ 0.821 ” 0.835“
(0.39) (0.39) (0.39) (0.39)
2 years of Schooling 0.482‘” 0.462‘” 0.320 0.313
(025) (025) (0.25) (0.25)
3 years of Schooling 0.149 0.111 0.066 0.056
(0.19) (0.19) (0.19) (0.20)
4 years of Schooling 0280 0.231 0.215 0.211
(0.22) (0.22) (0.22) (0.23)
5 years of Schooling - 0.025 - 0.073 - 0.131 - 0.144
(0.25) (0.25) (0.25) (0.26)
6 years of Schooling - 0.058 - 0.095 - 0.053 - 0.030
(0.10) (0.10) (0.1 1) (0.12)
7 years of Schooling 0.045 0.040 0.099 0.172
(0.43) (0.43) (0.43) (0.43)
8 years of Schooling - 0.309 - 0.383 - 0.280 - 0.233
(0.31) (0.31) (0.32) (0.32)
9 years of Schooling - 0.352 - 0.403‘” - 0.243 - 0.172
(024) (024) (0.24) (0.25)
10 years of Schooling - 0.694‘” - 0.777” - 0.636‘” - 0.499
(0.36) (0.35) (0.35) (0.36)
11 or more years of - 1264‘ - 1.337‘ - 1.250‘ - 1.076‘
Schooling (0.13) (0.13) (0.15) (0.17)
Cohort Effects
18-22
23-27
28-32 0.321 " 0.362' 0.354'
(0.12) (0.12) (0.12)
33-37 0502' 0.525‘ 0.513‘
(0.13) (0.13) (0.12)
38-42 0.087 0.126 0.122
(0.13) (0.13) (0.13)
Community Variables
Urban Residence - 0.302‘ - 0.296‘
(0.08) (0.09)
Protestant 0.164 0.154
(0.15) (0.15)
Catholic 0.255 0.239
(0.16) (0.16)
lslarn - 0228 - 0.231
(0.17) (0.17)
Polygamy - 0.129 - 0.135
(0.09) (0.09)
Husband's Education
1-5 years of Schooling 0.138
(0.15)
6 years of Schooling 0.042
(0.1 1)
7-10 years of Schooling 0.184
(0.18)
11+ years ofschooling - 0.216‘“
(0.13)
intercept 3.586‘ 3.349‘ 3.410‘ 4.405‘
(0.06) (0.1 l) (0.17) (0.17)
R-squared 0.05 0.06 0.08 0.08
F( all Covariates) . 10.98 10.16 9.93 8.52
N 2197 2197 2197 2197

 

146

TABLE 6.9 - Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husbands Education ( Ever-married at Age 33)

 

 

 

 

 

 

Dependent Vanab' 1e: Cumulative fertility At Age 33
Reg 1 Reg 2 Reg 3 Reg 4
lumident Variables:
1 year of Schooling 1.457” 1.446" 1.374‘ 1.362‘
(0.53) (0.53) (0.53) (0.53)
2 years of Schooling 0.307 0.291 0.071 0.051
(0.34) (0.34) (0.34) (0.34)
3 years of Schooling 0.742" 0.697” 0.661 “ 0.623”
(0.28) (0.28) (0.28) 0.28)
4 years of Schooling 0.536‘ 0.474 0.535‘” 0.512‘”
_ (0.31) 0.31) (0.31) (0.31)
5 years of Schooling 0.184 0.133 0.129 0.1 17
(0.37) (0.36) (0.37) (0.37)
6 years of Schooling 0.102 0.071 0.234 0.246
(0.15) (0.15) (0.17) (0.18)
7 years of Schooling 0.863 0.869 1.059 1.1 13‘“
(0.65) (0.66) (0.67) (0.65)
8 years of Schooling - 0.586 - 656 - 0.345 - 0.277
(0.50) (0.50) (0.51) (0.52)
9 years of Schooling - 0.355 - 0.409 - 0.130 - 0.075
(0.38) (0.37) (0.53) (0.39)
10 years of Schooling - 0.778 - 0.883 - 0.578 - 0.474
(0.53) (0.53) (0.53) (0.54)
11 or more years of -1.321‘ - 1.403‘ - 1.160‘ -1.048‘
Schooling (0.22) (0.22) (0.24) (0.27)
Cohort Effects
18-22
23-27
28-32
33-37 0475‘ 0.507‘ 0.496‘
(0.15) (0.15) (0.15)
38-42 0.158 0.215 0.222
().15) (0.15) (0.15)
Community Variables
Urban Residence - 0.521‘ - 0.514’
(0.13) (0.14)
Protestant 0.01 l - 0.025
(021) (0.22)
Catholic 0. 130 0.023
(013) (023)
Islam - 0.482“ - 0.498”
(0.24) (0.24)
Polygamy - 0.181 - 0.190
(0.12) (0.12)
Husband's Education
1-5 years of Schooling 0.309
(0.20)
6 years of Schooling 0.066
(0.16)
7-10 years of Schooling 0.083
(0.26)
1 1+ years of schooling - 0.093
(0.21)
Intercept 1.955‘ 4.733‘ 5.011‘ 4.992'
(0.08) (0.13) (0.22) (022)
R-squared 0.05 0.05 0.08 0.08
F( all Covariates) 6.10 6.13 6.90 5.79
N 1427 1427 1427 1427

 

TABLE 6.10 - Impact of Education on Fertility - Female Schooling, Cohort Effects,
Community Variables and Husbands Education ( Ever-married at Age 38)

 

 

 

 

 

 

 

Dependent Variable: Cumulative fertility At Age 38
Reg 1 Reg 2 Reg 3 Reg 4
Independent Variables:
1 year of Schooling 1.629“ 1.631” 1.780” 1.693”
(0.76) (0.76) (0.74) (0.74)
2 years of Schooling 0.472 - 0.424 0.131 0.003
(0.51) (0.51) (0.50) (0.50)
3 years of Schooling 1.018“ 1.030“ 1.027‘ 0.918“
(0.41) 0.41) (0.41) (0.41)
4 years of Schooling 0.599 0.608 0.848’” 0.850'"
(0.40) (0.40) (0.48) (0.48)
5 years of Schooling - 0.166 - 0.167 - 0.043 - 0.117
(0.58) (0.50) (0.57) (0.58)
6 years of Schooling - 0.027 - 0.027 0.353 0.357
(0.23) (0.23) (0.24) (0.26)
7 years of Schooling 0.992 1.003 1.503‘” 1.580‘”
(0.88) (0.69) (0.87) (0.88)
8 years of Schooling - 0.452 - 0.455 0.121 0.201
0.83) 0.83) (0.82) (0.85)
9 years of Schooling - 0.640 - 0.633 - 0.103 - 0.062
(0.58) (0.58) (0.57) (0.59)
10 years of Schooling - 0.437 - 0.428 0.112 0.218
0.94) (0.94) (0.93) (0.96)
11 or more years of - 1.508‘ - 1.503‘ - 0.951‘ - 0.840”
Schooling (0.36) (0.36) (0.38) (0.42)
Cohort Effects
18-22 '
23-27
28-32
33-37
38-42 0.070 0.010 0.036
‘ (0.17) (0.17) (0.17)
Community Variables
Urban Residence - 0.992‘ - 1.015‘
(0.19) (0.19)
Protestant - 0.148 - 0.208
(0.30) (0.30)
Catholic - 0.280 - 0.347
(0.33) (0.33)
lslarn ,- 0.733“ 0.737“
(0.33) (0.34)
Polygamy - 0.226 - 0.240
(0.18) (0.17)
Husband's Education
1-5 years of Schooling 0.850‘
(0.30)
6 years of Schooling 0.134
(0.23)
7-10 years of Schooling 0.096
" (0.40)
1 1+ years of schooling 0.016
(0.30)
Intercept 6.01 ‘ 6.049‘ 6.669‘ 6.597‘
(0.1 l) (0.15) (0.29) (0.29)
R-squared 0.04 0.04 0.09 0.10
F( all Covariates) 3.27 3.01 5.03 4.479
N 875 875 875 875

 

6.4.6 URBAN AND RURAL SAMPLES

In Tables A63 and Table A6.4 we have the regression result for women living in
the urban and rural areas in the South for our third set of equations. As evident in the
tables, fertility declines with higher levels of female schooling in both the rural and urban
areas. At age 18 in particular, cumulative fertility declines with two or more years of
schooling for women in the rural areas but only with six or more years of schooling for
women in the urban areas. This is consistent with empirical work in the literature which
suggests that women with lower levels of schooling may have slightly higher fertility in
the urban areas than the rural areas due to greater relaxation of traditional fertility
restraints in urban areas than rural areas (Singh and Casterline, 1985). In both the urban
and rural areas we observe an inverted U-shaped distribution of births among the age
cohort dummies from the oldest to the youngest cohort. However, within each age group
women residing in the rural areas have born more children than those in the urban areas.
Among the community variables, however, has predictive power in the urban areas but
Islam has a significant negative influence on fertility in the rural areas.

We make similar observations for cumulative fertility at age 23. In the rural areas
we have scattered influence of female schooling on cumulative fertility at age 23, on the
other hand women with eight or more years of schooling in the urban areas have fewer
children. Community variables do not make any impact in both the urban and rural areas.

At age 28, however, the influence of schooling diminishes in both the urban and

rural areas. In the urban areas it is only women with eleven or more years of schooling

149

who have fewer children, but in the rural areas women with nine and eleven or more
years of schooling have fewer number of children born at age 28.

At higher attained ages, influence of female schooling diminishes still further in
both the rural and urban areas. In both areas it is only women with eleven or more years
of schooling who have fewer children at age 33 and 38. The influence of eleven or more
years of schooling however, has a greater impact on fertility for women in the rural areas
than the urban areas. None of the community variables has influence on cumulative
fertility at ages 33 and 38 in the rural areas but Islam has a significant negative impact on
cumulative fertility at ages 33 and 38 in the urban areas. At age 33 and 38, however
women ages 33-37 and 38-42- in the urban areas have more children than those in the
rural areas.

We tested if the rural and urban regressions were significantly different and found
them to be jointly significant. The early years of schooling is sometimes associated with
higher fertility in the urban areas at all attained ages though its influence is not
significant. On the other hand in the rural areas women with one year of schooling have

Significantly more children born at all attained ages.

6.5. SUMMARY

 

In this chapter, we have analyzed the relationship between female schooling,
relative cohort position and community variables on such key correlates of fertility such
as cumulative fertility, age at first maniage and age at first birth in southern Nigeria

where it is believed that fertility had fallen in recent years.

150

Our results indicate that lower levels of female schooling does not influence
fertility decision making. It is only after a certain threshold has been reached that
increases in female schooling begin to have a negative impact on fertility. Schooling has
a differential impact on fertility depending on a woman’s age. Among the youngest
women 18-27, the threshold is about four or more years of schooling, among women
ages 28-37 the threshold is about six or more years of schooling, and with the oldest
women 38-47 it is eight or more years of schooling. We make similar observations for
the relationship between female schooling and age at fast marriage and age at first births
We observe that about 25 percent to 61 percent of the decline in fertility is accounted for
by female schooling, controlling for the relative cohort position of the woman.

Urban residence has a significant negative influence on fertility and age at first
marriage and age at first birth. On the other hand, Protestant or Catholic background
often has unpredictable effect, whilst that of Islam background makes a significant
negative impact on cumulative fertility and the other proximate determinants of fertility.
The inclusion of women in polygamous unions did not change our results, in addition it
was not significant in all of our regressions. .

Among the ever-married women, we observed that within each cohort the ever-
married have born more children than women in the all sample. Husband’s education
actually raises fertility for younger women, whilst its impact is negligible for the older
women. It is only women whose husbands have either six or eleven or more years of
schooling that have sometimes fewer children at age 18 and higher attained ages

respectively.

151

In looking at the urban and rural women we observe that higher levels of female
schooling makes a significant negative impact on fertility in both the rural and urban
areas, with its greatest impact in the rural areas. Islamic background makes a significant
impact on fertility in rural areas at younger ages and in the urban areas at higher
attained ages. Protestant and Catholic backgrounds, however, make some impact in both
the rural and urban areas.

We had anticipated that Islamic background would have a positive impact on
fertility and other proximate determinants of fertility but our results point to the contrary.
This may be in part because of the restriction of our regression analysis to women in the
south since Islam is practiced on a wider scale in the North. It is also possible that
women who are in the Islamic faith in the south are better educated migrants fi'om the
North. We have similar reasons for the insignificant impact of polygamous union on

cumulative fertility for the ever-married women.

152

CHAPTER 7

INCREASES IN WOMEN'S EDUCATION AND FERTILITY DECLINE

7.1. INTRODUCTION

 

From our analysis in chapter 5, there was little or no evidence to suggest an
overall fertility decline in Nigeria in spite of the tremendous increases in schooling for
women. There was some evidence however, of a decline in fertility in the Southwest
and the Southeast among the youngest two birth cohorts. In chapter 6 we also examined
in some detail the relationship between female schooling and cumulative fertility at
various attained ages and other proximate determinants of fertility. Our analysis suggests
that women who have completed primary school and beyond consistently have fewer
children ever born by ages 18, 23 and 28. At higher attained ages 33 and 38, however,
the influence of schooling on fertility diminishes, but that of eleven or more years of
schooling persists. Similar observations were made for first marriages and first births at
various attained ages.

A sustainable fertility transition must be accompanied by younger women having
fewer children, marrying later and having their first births later than the older women at
each of the various attained ages. However, even from the regressions in chapter 6, we
still observe an inverted U-shaped distribution of children ever born from the oldest to the
youngest women. In addition we found evidence to support that younger women do first

marry and have their first births earlier than the older women. Evidence from chapter 6

153

suggests that women with higher levels of schooling have fewer children but still we may
not know from the analysis that fertility has fallen.

In this chapter, we investigate further to see if fertility has indeed fallen and if so
among which socio-economic groups', using female schooling as our measure of social
stratification. We run a series of regressions in this chapter over different levels of
female schooling. Our object is to see if women of the same age group behave differently
depending on their educational status with regard to fertility decision making. We report
the results for cumulative fertility at ages 18, 23 and 28 since at higher attained ages we
have fewer age groups to compare and the regressions coefficient were not jointly
significant. For similar reasons and for the fact that at higher attained ages we have
' almost universal first maniages and first births we report that for first marriages and first
births at ages 18 and 23.

The rest of the chapter is organized as follows: Section II first looks at the trend
' in cumulative fertility over different schooling levels and second at the multivariate least
squares analysis of fertility over different schooling levels. Section III also discusses the
logit regressions of age at first marriage and age at first births over different levels of

female schooling. Section IV concludes the chapter.

7.2. Increases In Women's Schoolingand Fertility Decline

 

7.2.1 Cumulative Fertility at Various Attained Ages by Years of Schooling
In Figure 7.1 A through Figure 7.2D we look at the trend in cumulative fertility

among age groups of women in the survey over different levels of female schooling. As

154

Figure 7.1A shows at age 18, the cumulative fertility profiles for women with either no
schooling or 1-5 years of schooling rise steadily fi'om the oldest women reaching their
peaks with women ages 23-27 and then declines with the youngest women.. On the
other hand , for women with six years of schooling the profile remains fairly constant for
women ages 33-47 but thereafter rises steadily fi'om the oldest to the youngest women. In
sharp contrast, that of women with seven or more years of schooling rises initially but
then declines steadily from the oldest to the youngest women.

In Figure 7.1B we make similar observations for cumulative fertility at age 23.
Cumulative fertility rises steadily from the oldest to the youngest women with no
schooling or 1-5 years of schooling. That of women with six years of schooling
declines with the two youngest age cohorts. On the other hand, cumulative fertility of
women with seven or more years of schooling declines steadily from the oldest to the
youngest age cohort.

At age 28 as Figure 7.lC shows, we observe a more pronounced inverted U-
> shaped distribution for women with no schooling or 1-5 years of schooling but that of
women with six years of schooling rises steadily from the oldest to the youngest women.
On the other hand, the profile for women with seven or more years of schooling declines
steadily from the oldest women to the youngest women. At age 33 on the other hand ,
cumulative fertility for women with no schooling, 1-5 years and six years of schooling all
rise uniformly from the oldest to the youngest women. In sharp contrast, however, the
cumulative fertility for women with seven or more years of schooling declines steadily

from the oldest to the youngest women as evident in Figure 7.1D.

155

Several implications can be drawn from the analysis presented in Figures 7.1A
through 7.1D. First, since cumulative fertility at age 18 for all levels of schooling and
for all the ages groups is less than 1.00, this means Nigerian women have started
motherhood at age 18 but have not started parenting yet. Women with no schooling
report the highest cumulative fertility, followed by those with 1-5 years of schooling,
followed by women with six years of schooling, women with seven or more years of
schooling reporting the lowest cumulative births at age 18. There is therefore, a clear
inverse relationship between schooling and cumulative fertility at age 18.

At age 23, however, the cumulative fertility for women with no schooling, 1-5
years and six years are almost indistinguishable and all women have started parenthood.
On the other hand, women with seven or more years of schooling report the lowest
cumulative fertility and the last two youngest women have not started parenthood at age
23.. However, even at age 23 the inverse relation between education and fertility remains
for the two youngest women. At ages 28 and 33 however, the perfect inverse
relationship between cumulative fertility and fertility disappears since women with 1-5
years and six years of schooling have more children at ages 28 and 33 than women with
no schooling. It is women with seven years of schooling who have the lowest fertility
even at ages 28 and 33. We can conclude that women 1-5 years and six years of
schooling are the most likely to postpone births at ages 18 and 23 but then catch-up by

age 28 and above.

156

Figures 7.2A through 7.2D are another look at the same data but this time for
cumulative fertility at various attained ages for different levels of schooling. As evident
in Figure 72A for women with no schooling cumulative fertility at various attained ages
rises steadily from the oldest to the youngest women. We begin to observe an inverted
U-shaped distribution for women with 1-5 years and six years of schooling for
cumulative fertility at ages 23, 28 and 33. In sharp contrast, for women with seven or
more years of schooling, cumulative fertility at all attained ages decline from the oldest to
the youngest women.

The evidence above suggests that cumulative fertility may have fallen only for
women with seven or more years of schooling, the better educated women. However, it
may also be better reporting of births by the more educated women relative to the least
educated. In the next section we run a series of regressions to investigate whether fertility

has indeed fallen and for women in which schooling category.

7.2.2 Education and Fertility: Results by Levels of Female Schooling

Table 7.1 presents our regression analysis for cumulative fertility at age 18 for
women with no schooling, 1-6 years of schooling and seven or more years of schooling.
We combined women with 1-5 years and six years of schooling because the slopes of the
profiles were very similar.

We begin by running our first set of regressions with only cohort dummies and
then add education and later community variables. For women with no schooling we

have only two specifications. A look at cumulative fertility at age 18 for women with no

157

schooling reveals that relative to the oldest women ages 43-47, all the younger women
have more children except those ages 18-22. Children ever born at age 18 rises steadily
with the 38-42 age group to significant high of 0.17 and 0.18 with women ages 28-32 and
23-27 respectively, a pattern very similar to the inverted U-shaped distribution discussed
above. We make similar observation in our second equations for women with no
schooling. Urban residence and Islam have a negative influence on fertility for women
with no schooling.

Among women with 1-6 years of schooling we continue to observe that relative
to the oldest women, all the younger women have borne more children at age 18. The
result is significantly higher for women ages 33-37 in our first specification and for
women ages 23-27, 28-32 and 33-37 in our second specification. Relative to women with
one year of schooling women with 2-6 years of schooling all have fewer children, an
observation we made in both our second and third specifications.

The picture is significantly different for women with seven or more years of
schooling. Relative to the oldest women ages 43-47, the last three youngest age cohorts
all have fewer children even thOugh the result is not significant, an observation we also
‘ made in all specifications. None of the community variables has any predictive influence
on cumulative fertility at age 18. Relative to the women with seven years of schooling
those with eight or more years of schooling have fewer children.

As Table 7.2 shows, at age 23 for women with no schooling women ages 23-27
and 28-32 have significantly more children. Cumulative fertility at age 23 increases

steadily from the oldest to the youngest women. This is an observation we make when

158

we also control for community variables. Urban residence depresses fertility, so does
Islamic influence. For women in with 1-6 years of schooling, in all three specifications
all younger women have more children born at age 23 than the oldest women with the
women of ages 33-37 being significantly higher. Relative to women with one year of
schooling , those with three or more years of schooling have fewer children in our second
set of equations. We make similar observations when we control for community
variables. Women with Protestant or Catholic backgrounds have more children .

As we get to women with seven or more years of schooling we make remarkably
different observations. In all of our three sets of equations, the younger women have
fewer children than the oldest women, with that of the youngest women ages 18-22 being
significantly lower in two of the specifications. At age 23 however, it is only women with
ten or more years of schooling who have fewer number of children in both specification.
In our third specification none of the community variables has predictive power.

At age 28 and among women with no schooling we again observe that women
ages 28-32 and 33-37 have significantly more children as evident in Table 7.3. Urban
' residence makes a significant negative impact on cumulative fertility at age 28 and so is
Islamic affiliation. We make similar observations with women who have 1-6 years of
schooling. Relative to women with one year of schooling women with 2-6 years of
schooling have fewer children in both specifications that female education is controlled
for. on the other hand, among women with seven or more years of schooling, we
observe that younger women have fewer children with that of the 28-32 being

significantly lower. We observe that relative to those with exactly seven years of

159

schooling, it is only women with ten or more years of schooling who have significantly
fewer children.

In sum, in looking at the regressions for women in different schooling categories,
we observe that for women with no schooling or 1-6 years of schooling the younger
women have more children than the oldest women, an observation we make for

cumulative fertility at ages 18, 23, and 28 respectively.

7.3.3 Age at First Marriage and Age at First Birth: By Levels of Female Schooling

In this section we also run logit regressions for age at first marriage and at first
birth over levels of female schooling to see if women in the same age group behave
differently depending on their educational status with regard to marital and birth decision
making. We report the results for only first marriages and first births at ages 18 and 23

since there is almost a universal marriage and first births at ages 28 and over.

Age at First Marriage

.Table 7.4 reports first marriages at age 18 for women with no schooling, 1-6 years
of schooling and seven or more years of schooling. For women with no schooling in our
first regression we observe that women ages 23-27 and 28-32 significantly marry early
relative to the oldest women. When we control for community variables the magnitude
and significance of the cohort dummies do not change very much. In addition urban
residence delays first marriages but Catholic background ushers women into first

marriages at age 18 early.

160

For women with 1-6 years of schooling, we observe that relative to the oldest
women, the younger women (except those ages 33-3 7) marry early even though the
coefficients are not significant. In addition relative to women with one year of schooling
those with 2-6 years of schooling do marry late at age 18. Urban residence delays first
marriages at age 18 whilst women with 2-6 years of schooling all marry late relative to
those with one year of schooling.

As we move to women with seven or more years of schooling we observe that the
younger women postpone marriages at age 18 as compared to the oldest women.
Women ages 18-22 and 23-27 significantly postpone their marriages at age 18 than the
oldest women. We make similar observations when we control for years of schooling.
Women in all the schooling categories marry late relative to those with exactly seven
years of schooling, with that of women with eight and ten years of schooling being
significant. In the third set ‘of regressions when we control for community variables as
well, we observe that women in the urban areas as well as women in all the religious
groups marry late at age 18.

We make a similar observation for first maniages at age 23. Urban residence
delays first marriages at age 23, but religious influence ushers women into early first
marriages at age 23. For women with 1-6 years of schooling we observe that relative to
the oldest women all the younger women first marry early. None of the community
variables make any significant impact on the timing of first marriages at age 23.

We make different observations for women with seven or more years of

schooling. In all the three specifications the younger women many late relative to the

161

oldest women even though the coefficients are not significant. It is only urban residence

and eleven or more years of schooling that have a delay impact on first marriages at age

23.

Age at First Birth

In Table 7.6 we document first births at age 18 over levels of female schooling.
In both specifications for women with no schOoling, we observe that younger women
have their first births early at age 18 than the oldest women with that of women ages 23-
27 and 28-32 being significant. Urban residence and Islamic influence delays the onset
of first births at age 18. For women with 1-6 years of schooling we again observe that
relative to the oldest women younger women have their first births early, the results being
significant for women ages 28-32 and 33-3 7. In addition relative to women with one
year of schooling, women with four and six year of schooling delay their first births at
age 18. In the third specification none of the community variables has a significant
impact on the timing of first births at age 18.

For women with seven or more years of schooling, we make slightly different
observations. In all the equations the three youngest women delay the onset of their first
births even though the coefficients are not significant. Relative to women with seven
years of schooling, women with eight, ten and eleven or more years of schooling delay
their onset of first births. All the community variables delay the onset of first births with

that of urban residence being significant.

162

At age 23 as Table 7.7 shows, for women with no schooling, the youngest two
cohorts have their first births at age 23 early relative to the oldest women. Urban
residence delays the onset of first births at age 23, but Protestant and Catholic
background quicken the onset of first births at age 23.. For women with 1-6 years of
schooling, we also observe that relative to the oldest women all the younger women have
their first births early with that of women ages 33-37 being significant. This is an
observation we make in all the three specifications. None of the community variables
make any significant impact on first births. Apart from women with three years of
schooling who delay their first births, the rest are not significant. In sharp contrast for
women with seven or more years of schooling, we observe that relative to the oldest
women. the younger women have their first births late with that of women ages 23-27
being significant in the first equation. None of the community variables has predictive
influence of first births at age 23. Relative to women with seven years of schooling only
women with eleven or more delay their first births at age 23.

The effect of female schooling on cumulative fertility, age at first marriage and
age at first births are very similar. At lower levels of schooling, the younger women have
more children, marry early and have their first births early relative to the oldest women.
However for women with seven or more years of schooling, we observe that the younger
women have fewer children, first marry late and delay the onset of first births. This

suggests then that fertility may have fallen among the better educated women.

163

 

7.4. SUMMARY

The main objective of this chapter was to investigate if women behave differently
with regards to fertility decision making depending on the educational level. Our second
object was also to find out if fertility has fallen and if so among which socio-economic
group. The result indicates that fertility may have fallen among the better educated
women in the south. Among women with no schooling and 1-6 years of schooling we
observe that the younger women rather have more children relative to the oldest women.
However, among women with seven or more years of schooling, we observe that
younger women have fewer children. An observation we made for cumulative fertility at
ages 18, 23, and 28 and in all our three specifications.

We make similar observations for other proximate determinants of fertility such
as age at first marriage and age at first birth. Among women with no schooling and 1-6
_ years of schooling the younger women are marrying early and having their first births
, earlier than the oldest women. On the contrary among women with seven or more years

of schooling, we observe that the younger women delay their first marriages and the on

set of their first births.

Cumulative Fertility

1
0.9

 

 

 

 

 

 

 

 

 

 

 

 

 

g 0.7
3 0.6 7/ /\
.436 05 /—-—/ \e
'5 e
g 0.4 // i c\ c ‘f—
0 0.3 3f ,
A
it \~\
0.1 _ _.
o I T I I I
43.47 3842 8387 2842 28-27 18-22

Age Group

 

+0years -o—1-6 years +6years +7+ years]

Figure 7.1A- Cumulative Fertility at Age 18

 

 

 

 

 

 

 

 

 

 

 

by Years of Schooling.

2.5

2 M4

\

1 .6

1 ; \\
0.6

O r . r i

4347 38-42 33-37 28-32 23-27

Age Group

 

l—o—Oyeara —o—1-Syeara —-—Cyeara +7+ yearn—l

Figure 7.lB- Cumulative Fertility at Age 23
by Years of Schooling

I65

Cumulative Fertility

Cumulative Fertility

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4
3.5 -

3
2.5

2
1.5

1
0.5

0 r r r . a

4347 38-42 33-37 28-32 23-27

Age Group
+ 0 years +13 years -— 6 years -a— 7+ years
Figure 7.1C- Cumulative Fertility at Age 28
by Years of Schooling
B ,__.
4 __ — ¢
3
2
1
o , . . s
4347 3842 33-37 28-32 23-27
Age Group

 

[—o—Oyears +1-5yeara +6yeara +7+ years]

 

Figure 7.1D- Cumulative Fertility at Age 33
by Years of Schooling

165A

Cumulative Fertility

Cumulative Fertility

 

 

 

 

 

 

 

 

 

 

 

 

 

5 M

4

3 /

2 j 4-
A/

1 4 .
1‘ + - fi

0 , I I I I 1

43-47 38-42 33-37 28-32 23-27 18-22
Age Group

 

+ cumf18 -o— cumf23 + cumf28 -n— cumf33

 

 

 

Figure 7.2A-Cumulative Fertility at various
Attained Ages by Years of Schooling (0 years)

 

 

 

 

 

 

 

 

 

 

 

6

5 //

4 ‘— A

3 ‘/

2 1% f A
tr/

‘1

o l?" a - ‘5

 

43-47 38-42 33-37 28-32 23-27 18-22
Age Group

 

—a— cumft 8 -o- cumf23 -— cumf28 -— cumf33

 

Figure 7.28- Cumulative Fertility at various Attained
Ages by Years of Schooling (1-5 years)

166

Cumulative Fertility

Cumulative Fertillty

 

 

 

N

 

 

.L

 

 

 

f—————m——-—-——“ F —r a
01

4347 3842 3337 28-32 23-27 18-22
Age Group

 

F" cumfl 8 + cumf23 -0— cumf28 -I- cumf33]

Figure 7.2C- Cumulative Fertility at various
Attained Ages by Years of Schooling (6 years)

5
4.5
4
3.5
3
2.5
2
1 .5
1
0.5
0
4347 3342 33-37 2842 23-27 1 8-22

Age Group

 

 

+ cumt18 + cumf23 + cumf28 + cumf33 ]

Figure 7.20- Cumulative Fertility at various Attained
Ages by Years of Schooling (7+ years)

I66A

TABLE 7.1 - Impact of Education on Fertility - Female Schooling, Cohort Effects
, and Community Variables by level of Female Schooling

 

 

 

 

 

 

 

Dependent Variable: Cumulative fertility At Age 18
Woman A55: 1847
0 Years of Schoolin 1-6 Years of Schooling 7+ Years of Schooling
Reg 1 Reg Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
Independent Variables:
Cohort Effects
18-22 - 0.003 0.027 0.045 0.079 0.075 - 0.107 - 0.088 - 0.097
(0.1 1) (0.1 l) (0.08) (0.08) (0.09) (0.07) (0.07) (0.07)
23-27 0.181’” 0.217” 0.115 0.151‘“ 0.151‘” - 0.093 - 0.066 - 0.077
(0.10) (0.10) (0.09) (0.08) (0.09) (0.07) (0.07) (0.07)
28-32 0.172” 0.204‘ 0.126 0.141‘" 0.141‘” - 0.039 - 0.020 - 0.020
(0.08) (0.08) (0.09) (0.09) (0.09) (0.07) (0.08) (0.08)
33-37 0.010 - 0.001 0.204” 0.202” 0.199“ 0.005 0.010 0.016
(0.09) (0.09) (0.09) (0.09) 0.09) (0.08) (0.08) (0.08)
38-42 0.037 0.062 0.071 0.058 0.058 0.056 0.049 0.056
(0.08) (0.09) (0.09) (0.09) (0.09) (0.09) (0.08) (0.09)
Community Variables
Urban Residence - 0.153" - 0.028 - 0.103‘
(0.06) (0.04) (0.02)
Protestant - 0.076 0.158 0.018
(0.07) (0.11) (0.1 1)
Catholic 0.022 0.194‘” - 0.003
(0.08) (0.11) (0.11)
Islam - 0.278‘ 0.140 0.003
(0.09) (0.12) ().l 1)
Education
2 years of Schooling 0.470‘ - 0.473”
(0.16) (0.0.17)
3 years of schooling - 0.452' - 0.448“
(0.16) (0. l6)
4 years of schooling - 0.478‘ - 0.472‘
(0.16) (0.16)
5 years of schooling - 0.640‘ - 0.645‘
(0.16) (0.17)
6 years of schooling - 0.610‘ - 0.603‘
(0.15) (0.15)
8 years of schooling - 0.232‘ . 0.226‘
(0.06) (0.06)
9 years ofschooling 0.122“ - 0.118‘”
(0.06) (0.06)
10 years of schooling - 0.207‘ - 0.281 ‘
(0.06) (0.06)
1 1 or more years of -0.285‘ - 0.268‘
schooling (0.06) 90.06)
intercept 0.539‘ 0.661‘ 0.308‘ 0.855‘ 0.707‘ 0.194“ - 0.422‘ 0.485‘
(0.06). (0.09). (0.07) (0. l 6) (0.19) (0.07) (0.09) (0.13)
R-squared 0.01 0.04 0.01 0.02 0.03 0.01 0.04 0.06
F( all Covariates) 2.00 5.37 1.66 3.33 2.65 3.55 8.02 7.07
F(Community) - 9.52 - - 0.97 - - 4.78
F(Schooling) - - - 4.97 - 13.34 -
N 1326 1326 1436 1436 1436 1555 1555 1555

 

Note: Standard errors in parenthesis> ‘np-value < 0.01, "- p-value < 0.05, "‘u p-value < 0.10..

167

TABLE 7.2 - Impact of Education on Fertility - Female Schooling, Cohort fleets,
and Community Variables by level of Female Schooling

 

 

 

 

 

 

Dependent Variable: Cumulative fertility At Age 23
Woman Ages 23-47
0 Years of Schoolin I - 6 ears of Schoolin 7 + Years of Schoolin
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
Independent Variables:
Cohort Effects
18-22
23-27 0.339” 0.409” 0.069 0.143 0.141 - 0.515“ 0.387” - 0.391“
(0.18) (0.14) (0.19) (0.17) (0.170 (0.21) (0.21) (0.21)
28-32 0.336” 0.397' 0.210 0.246 0.247 - 0.349 - 0.244 - 0.235
(0.14) (0.14) (0.17) (0.17) (0.17) (0.22) (0.21 ) (0.21 )
33-37 - 0.039 - 0.017 0.419” 0.417” 0.411“ - 0.154 - 0.099 - 0.082
(0.14) (0.14) (0.18) (0.18) (0.18) (0.24) (0.23) (0.23)
38-42 - 0.158 - 0.113 0.075 0.067 0.066 - 0.119 - 0.127 - 0.099
(0.14) (0.13) (0.18) (0.18) (0.18) (0.35) (0.24) (024)
Community Variables
Urban Residence - 0.274” - 0.074 - 0.122
(0.10) (0.09) (0.10)
Protestant 0.089 0.446‘” - 0287
(0.12) (0.24) (0.37)
Catholic 0.238‘" 0.529“ 0.120
(0.14) (0.25) (0.38)
Islam . 0.308” 0.431 - 0.244
(0.15) (0.26) (0.30)
Education
2 years of Schooling - 0.382 - 0.389
(0.34) (0.34)
3 years of schooling - 0.587‘“ - 0.581 ‘”
(0.32) (0.33)
4 years of schooling 0.551‘” - 0.537
(0.33) (0.33)
5 years of schooling - 0.768‘ - 0.779”
(0.34) (0.34)
6 years of schooling - 0.833’ - 0.818“
(0.30) (0.31)
8 years of schooling - 0.279 - 0.272
(0.26) (0.26)
9 years of schooling - 0.210 - 0.208
(0.25) (0.25)
10 years of schooling - 0.445‘” - 0.454’”
(0.27) (0.27)
11 or more years of - 0.963‘ - 0.947'
schooling (0.23) (0.23)
Intercept 1.848‘ 1.888‘ 1.654‘ 2.35‘ 1.933‘ 1.323 ‘ 1.956‘ 2.297‘
(0.1 l) (0.12) (0.15) (0.32) (0.39) (0.21) (0.29) (0.47)
R-squared 0.02 0.05 0.01 0.02 0.03 0.02 0.10 0.10
F( all Covariates) 5.54 7.51 2.48 2.89 2.39 4.30 12.52 8.61
F(Community) - 9.32 - - l .25 - - 0.81
F(Schooling) - - - 3.20 - - 20.29 -
N 1216 1216 1107 1107 1107 900 900 900

 

Note: Standard errors in parenthesis. ‘up-value <0.01, ”u p—value < 0.05 ”‘-o p—value < 0.10

168

TABLE 7.3 - Impact of Education on Fertility - Female Schooling, Cohort Effects ,
and Community Variables by level of Female Schooling

 

 

 

 

 

 

Dependent Variable: Cumulative fertility At Age 28
Woman Ages 28-47
0 Years of Schoolin l - 6 cars of Schoolin 7+ Years of Schoolin
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
Independent Variables:
Cohen Effects
18-22
23-27
28-32 0472" 0.563‘ 0.543" 0.590‘ 0.589" - 0.803” - 0.661“ - 0.633“
(0.18) (0.18) (0.21) (0.21) 0.21) (0.33) (0.32) (0.32)
33-37 0319‘” 0.359” 0.976‘ 0.901‘ 0.984‘ - 0.300 0.214 - 0.194
(0.19) (0.18) (0.22) 0.22) (0.22) (0.35) (0.34) (0.34)
38-42 - 0.059 0.002 0.476 0.474“ 0.483” - 0.360 - 0.366 o 0.349
(0.18) (0.17) (0.23) 0.23) (0.23) (0.3 7) (0.36) (0.37)
Community Variables
Urban Residence - 0.408‘ - 0.244'" 0.1 15
(0.14) (0.15) (0.24)
Protestant 0.239 0.346 0.373
(0.18) ‘ (0.41) (0.98)
Catholic 0.256 0.401 0.636
(0.20) (0.42) (0.99)
Islam - 0.341 ”‘ 0.304 0.298
(0.21) (0.44) (0.99)
Education
2 years of Schooling - 0.530 - 0.586
(0.45) (0.46)
3 years of schooling - 0.947” - 0.797“
(0.43) (0.43)
4 years of schooling - 0.769‘” - 0.696
(0.44) (0.44)
5 years of schooling - 1.025” - 0.982“
(0.46) 0.46)
6 years of schooling - 1.069" - 0.969”
0.40) (0.40)
8 years of schooling - 0.441 - 0.454
(0.47) (0.47)
9 years of schooling - 0.581 - 0.579
(0.43) (0.44)
10 years of schooling - 0.941 “ - 0.947"
(0.49) (0.49)
Ilormoreyearsof -l.454‘ -l.486‘
schooling (0.40) (0.40)
Intercept 3.319‘ 3.329‘ 3 .009‘ 3.924' 3.639" 3 .032‘ 4.059‘ 3.554‘
(0.14) (0.20) (0.18) (0.12) (0.57) (0.31) (0.47) (1.05)
R-squared 0.01 0.04 0.02 0.03 0.04 0.03 0.10 0.10
F( all Covariates) 4.56 6.71 6.68 4.03 3.02 3.70 6.38 4.25
F(Community) - 8.23 - - 1.00 - - 0.56
F( Schooling) - -— - 2,4 1 - - 8.29 -
N 1066 1066 790 790 790 423 423 423

 

Note: Standard errors in parenthesis. I'--p—value < 0.01, "u p-value < 0.05, *"-- p—value < 0.10.

169

TABLE 7.4 - Impact of Education on Age at First Maariage - Female Schooling, Cohort

Effects, and Community Variables by level of Female Schooling (Logit)

 

 

 

 

 

 

 

Dependent Variable: First Marriage at Age 18
Woman Ages 18-47
0 Years of Schoolin 1 - 6 cars of Schoolin 7+ Years of Schoolin
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
independent Variables:
Cohort Effects
18-22 - 0.097 - 0.034 0.128 0.266 0.198 - 0.796‘” - 0.711‘” - 0.768‘”
(0.24) (0.24) (0.23) (0.23) (0.23) (0.41) (0.42) (0.42)
23-27 0.499" 0.593‘ 0.188 0.333 0.279 - 0.724'” - 0.548 - 0.632
(0.23) (0.23) (0.23) (0.23) (023) (0.41) (0.43) (0.43)
28-32 0343‘” 0.420‘ 0.285 0.353 0.349 - 0.449 - 0.314 - 0.346
(0.19) (0.19) (0.23) (0.23) (0.23) (0.43) (0.44) (0.44)
33-37 0.006 0.020 0.514” 0.534” 0.547“ 0.171 0.280 0.267
(0.19) (0.19) (0.24) (0.25) (0.25) (0.45) (0.46) (0.47)
38-42 - 0.169 - 0.123 0.130 0.123 0.145 0.105 0.083 0.081
(0.18) (0.19) (0.25) 0.25) (0.26) (0.48) (0.50) ( 0.50)
Community Variables
Urban Residence - 0.377' - 0.285” - 0.289‘”
(0.13) (0.12) (0.16)
Protestant 0.152 - 0.291 - 1.349“
(0.17) (0.29) (0.57)
Catholic 0.535” - 0.086 - 1.488“
(0.1(0 (0.30) (0.58)
Islam - 0.148 - 0.415 - 1.578‘
(0.19) (0.32) (0.59)
Education
2 years ofSchooling - 0.910‘” - 0.911‘”
(0.55) (0.55)
3 years of schooling - 1.079“ - 1.011”
0.53) (0.53)
4 years of schooling - 1.469” - 1.392“
0.53) (0.53)
5 years ofschooling - 1.192” - 1.169"
(0.54) (0.54)
6 years of schooling . 1.,669‘ - 1.525‘
0.51) (0.51)
8 years of schooling - 0.800” - 0.736“
(0.35) (0.35)
9 years of schooling - 0.448 - 0.347
(0.32) (0.32)
10 years of schooling - 1.075‘ - 1.011‘
(0.35) (0.360
11 or more years of - 1.556‘ - 1.449.
schooling (0.30) (0.30)
Intercept 0.316" 0.271 ‘ - 0.231' 1.175" 1.482’ - 0.894“ 0.164 1.746“
- (0.14) (0.19) (0.20) (0.53) (0.60) (0.40) (0.49) (0.74)
- 2 Log Likelihood 1769.28 1736.62 1983.55 1951.19 1936.89 1470.85 1411.08 1399.06
Goodness of Fit 1325.97 1326.97 1436.00 1436.84 1436.86 1554.82 1547.68 1548.84
N 1326 1326 1436 1436 1436 1555 1555 1555

 

Note: Standard errors in parenthesis. ‘--p—vaiue < 0.01, "- p-value < 0.05, ""-- p-value < 0.10.

170

TABLE 7.5 - Impact of Education on Age at First Maariage - Female Schooling, Cohort
Effects, and Community Variables by level of Female Schooling (Logit)

 

 

 

 

 

 

 

Dependent Variable: First Maniage at Age 23
Woman Ages 23-47
0 Years of Schoolin 1 - 6 cars of Schoolin 7+ Years of Schoolin
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
independent Variables:
Cohort Effects
18-22
23-27 0796‘” 0.846‘” 0.354 0.414 0.423 - 0.662 - 0.489 - 0.466
(0.45) (0.45) (0.30) (0.31) (0.31) (0.41) (0.43) (0.43)
28-32 0.113 0.151 0.670” 0.676” 0.667“ - 0.187 - 0.013 - 0.030
(0.31) (0.31) (0.32) 0.32) (0.32) (0.42) (0.44) (0.44)
33-37 - 0.118 - 0.116 0.581‘” 0.571 ‘“ 0.570’” - 0.032 0.054 0.009
(0.31) (0.31) (0.34) (0.34) (0.34) ().45) (0.47) (0.48)
38-42 - 0.431 - 0.419 0.744” 0.697‘” 0.676‘” - 0.177 - 0.195 - 0.216
0.29) (0.29) (0.36) (0.36) (0.36) (0.48) (0.50) (0.50)
Community Variables
Urban Residence - 0.414‘” 0.243 0.406“
(0.21) (0.20) (0.19)
Protestant 0.434‘” - 1.008 - 5.806
(0.26) (0.74) (7.03)
Catholic 0.685“ - 1.026 - 5.782
(0.32) (0.76) (7.04)
Islam 0.545‘“ - 0.951 - 5.393
0.31) (0.78) (7.04)
Education
2 years of Schooling - 5.338 - 5.292
(7.45) (7.44)
3 years of schooling - 4.570 - 4.560
(7.45) (7.44)
4 years of schooling - 5.324 - 5.340
(7.45) (7.44)
5 years ofschooling -5.129 -5.119
(7.45) (7.44)
6 years of schooling - 5.305 - 5.346
(7.45) (7.43)
8 years of schooling - 0.210 - 0.205
(0.70) (0.71)
9 years of schooling - 0.435 - 0.451
(0.66) (0.66)
10 years of schooling - 0.592 - 0.584
(0.69) (0.69)
11 or more years of - 1.739‘ - 1.772‘
schooling (0.63) (0.63)
Intercept 2.219‘ 1.921 ‘ 1.498‘ 6.694 7.576 0.894“ 2.107’ 7.525
(0.24) (0.30) (0.25) (7.45) (7.46) (0.40) (0.73) (7.07)
- 2 Log Likelihood 793.46 785.48 806.81 796.11 792.31 1184.97 1112.64 1093.35
Goodness of Fit 1215.69 1219.93 1 106.99 1078.76 1074.40 900.00 904.95 891.36
N ' 1216 1216 1107 1107 1107 900 900 900

 

Note: Standard errors in parenthesis. ‘--p-value < t 0.01, "- p-value < 0.05, ""- p=value < 0.10.

171

TABLE 7.6 - impact of Education on Age at First Birth - Female Schooling, Cohort Effects
and Community Variables by level of Female Schooling (Logit)

 

 

 

 

 

 

Dependent Variable: First Births at Age 18
Woman Age: I84 7
0 Years of Schooth 1-6 Years of Schooling 7+ Years of Schooling
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 fl Reg 1 Reg 2 Reg 3
independent Variables:
Cohort Effects
18-22 0.076 0.145 0.252 0.352 0.331 - 0.605 - 0.485 - 0.583
(0.25) (025) (0.27) (0.27) (0.27) (0.56) (0.57) (0.57)
23-27 0.555“ 0.641‘ 0.349 0.453‘” 0.441 - 0.596 - 0.415 - 0.519
(0.22) (0.22) (0.27) (0.27) (0.27) (0.56) (0.57) (0.58)
28-32 0524‘ 0.601 ‘ 0.480‘” 0.523‘” 0.522‘“ - 0.093 0.046 0.026
(0.19) (0.19) (0.27) (0.27) (0.27) (0.57) (0.59) (0. 59)
33-37 0.114 0.139 0.652” 0.661” 0.661“ 0.240 0.357 0.391
(0.20) (0.20) (0.28) 0.28) (0.28) (0.60) (0.61) (0.62)
3842 0.096 0.146 0.212 0.187‘ 0.188 0.443 0.420 0.457
(0.19) (0.19) (0.29) 0.30) (0.30) (0.62) (0.64) (0.64)
Community Variables
Urban Residence - 0.310” - 0.126 - 0.674‘
(0.14) (0.13) (0.20)
Protestant 0.107 0.293 - 0.216
(0.17) (0.33) (0.81)
Catholic 0.244 0.362 - 0.574
(0.19) (0.34) (0.83)
islam - 0.350‘” 0.294 - 0.365
(0.20) (0.60) (0.84)
Education
2 years of Schooling - 0.580 - 0.584
(0.45) (0.46)
3 years of schooling - 0.558 - 0.534
(0.43) (0.43)
4 years of schooling -0.877“ - 0.848‘“
(0.44) (0.44)
5 years of schooling 1.218‘ - 1.212‘
(0.46) (0.46)
6 years ofschooling -1.055‘ - 1.013“
(0.41) (0.41)
8 years of schooling ~ - 0.956“
0.998“ (0.44)
(0.44)
9 years of schooling - 0.328 - 0.288
(0.37) 0.38)
10 years of schooling - 1.436‘ - 1.306‘
(0.47) (0.48)
11 or more years of - 1.425‘ - 1.297‘
schooling (0.35) 0.36)
Intercept -0.715‘ - 0.679‘ - 1.25 i ‘ - 0.368 - 0.632 - 1.910‘ - 0.944 - 0.169‘"
(0.15) (0.20) (0.24) (0.45) (0.54) (0.54) (0.62) (0.89)
-2 log Likelihood 1749.68 1723.76 1719.76 1703.78 1701.56 943.1 1 909.17 895.68
Goodness of Fit 1325.98 1326.61 1435.99 1435.63 1435.44 1554.89 1529.62 1533.61
N 1326 1326 1436 1436 1436 1555 1555 1555

 

Note: Standard errors in parenthesis> ‘-- p—value < 0.01, “-- p-value < 0.05, "n p-value < 0.10.

172

TABLE 7.7 - Impact of Education on Age at First Births - Female Schooling, Cohort

Effects and Community Variables by level of Female Schooling(Logit)

 

 

 

 

 

 

Dependent Variable: First Births at Age 23
Woman Ages 23-47
0 Years of Schoolin 1 - 6 ears of Schoolin 7 + Years of Schoolin
Reg 1 Reg 2 Reg 1 Reg 2 Reg 3 Reg 1 Reg 2 Reg 3
Independent Variables:
Cohort Effects
18-22
23-27 0865‘ 0.960‘ 0.153 0.206 0.215 - 0.787” - 0.634 - 0.627
(0.30) (0.30) (0.25) (0.26) (0.26) (0.39) (0.40) (0.41)
28-32 0389‘” 0.465“ 0.267 0.291 0.281 - 0.378 - 0.221 - 0.227
(0.22) (0.22) (0.26) (0.26) (0.26) (0.34) (0.42) (0.42)
33-37 - 0.021 0.009 0.575” 0.582“ 0.580" 0.036 0.121 0.108
(0.22) (0.22) (0.28) (0.29) (0.29) (0.43) (0.45) (0.45)
38-42 - 0.262 - 0.223 0.415 0.410 0.397 - 0.347 - 0.390 - 0.396
(0.21) (0.21 ) (0.29) (0.29) (0.29) (0.45) (0.48) (0.48)
CommunitLVariables
Urban Residence - 0.311” 0.058 0.136
(0.16) (0.16) (0.18)
Protestant 0.565” 0.136 . 0.611
(0.20) (0.39) (0.76)
Catholic 0.622" 0.087 - 0.599
0.23) (0.40) (0.77)
Islam 0.203 0.389 - 0.519
(0.22) (0.43) (0.77)
Education
2 years of Schooling 0.029 0.032
(0.64) (0.63)
3 years of schooling - 0.137“ - 0.149
(0.60) (0.59)
4 years of schooling - 0.316 - 0.321
(0.60) (0.60)
5 years of schooling - 0.616 - 0.612
(0.61) (0.61)
6 years of schooling - 0399 - 0.446
(0.56) (0.56)
8 years of schooling - 0.240 - 0.231
' . (0.57) (0.58)
9 years of schooling - 0.203 - 0.275
(0.54) (0.55)
10 years of schooling - 0.315 - 0.305
(0.58) (0.58)
11 or more years of' - 1.578' - 1.576‘
schooling (0.52) (0.52)
Intercept 1.125‘ 0.805‘ 0.950‘ 1.275” 1.121 ‘” 0.598 1.640" 2.113”
(0.16) 0.22) (0.22) (0.58) (0.68) (0.38) (0.63) (0.98)
- 2 log Likelihood 1278.44 1261.36 1 176.84 1 172.66 1 170.26 1227.33 1 150.36 1148.80
Goodness of Fit 1215.79 1219.07 1106.99 1107.87 1109.60 899.99 901.75 902.89
N 1216 1216 1107 1107 1107 900 900 900

 

Note: Standard errors in parenthesis. ’- p -value < 0.01, "- p-value < 0.05, , "‘"n p-value < 0.10.

173

CHAPTER 8

SUMMARY, CONCLUSIONS AND IMPLICATIONS

8.1 Introduction

 

This chapter is divided into three parts. The first part is a summary of the
previous chapters of data analysis. The second part is a discussion of our findings in
relation to other findings of research in the area. The third part is conclusions,

implications and suggestions for further research.

8.2 Summary

 

In this study we have investigated the evidence for a fertility decline in Nigeria
and also the relationship between female schooling and fertility using the NDHS, 1990
data. We started by first looking at the validity and reliability of the data. Our evidence
suggests that respondents tend to heap their ages on either "0" or "5". The Whipple's
Index for age concentration is 224.6 which is very high by the United Nations standards.
However, the Whipple's Index for age at first marriage and age at first birth were 121.73
and 112.78 respectively, which was approximate by United Nations standards. There was
also evidence to support the fact that the oldest women in the survey have memory
problems with recalling of their own ages, that of their children and also the number of
dead live-births they have had.

Second. by looking at the trends in female schooling, the study provided evidence

to support a substantial increases in female schooling among the younger women. Our

174

cohort analysis of fertility in general without reference to schooling, using such indicators
like cumulative fertility at various attained ages, age at first birth and age at first marriage
did not reveal any decisive decline in fertility.

Reconstructing both births and marital histories from the NDHS, 1990 data, our
evidence suggest that cumulative fertility may have declined over the younger women at
ages 18 and 23, but the overall decline in fertility is not large enough to offset the high
fertility of the oldest cohort. It is also possible that the young women may catch-up later
which we can not tell from data. We witness young women postponing first marriages
and first births at ages 18 and 23 but most women are married and have their first births at
age 28. In our regional analysis we found some evidence of fertility decline in the
Southwest and the Southeast but not in the North and also the magnitude that is suggested
in the recent empirical literature.

A look at the relationship between the mean CEB and single years of female
schooling revealed an eventual negative relation but one that is far from linear. We
observe that female schooling has a differential impact on fertility depending on the
woman's age. Female schooling has a greater negative impact on fertility for women ages
' 28-37 and the mildest impact for women ages 38-47.

Our multivariate regression analysis provided evidence on differentials in fertility
and other correlates of fertility by levels of female schooling. There is a powerful
negative relation between female schooling beyond primary and fertility. Women who
have completed primary and beyond have significantly fewer number of children. They

also marry late and have their first births later than women who have no schooling. This

175

is an observation we made both without controlling for and also controlling for age
cohort and community variables. The first four or five years of schooling has either a
positive or no relation on cumulative fertility at ages 18 and 23. At higher attained ages
28, 33 and 38, it requires a higher threshold of female schooling to make a negative
impact on fertility. At ages 28, there is a powerful negative relationship between female
schooling and fertility only with nine or more years of schooling, however at ages 33 and
38 it is only women with eleven or more years of schooling who have fewer children.
Thus indicating that the influence of female schooling on fertility diminishes at higher
attained ages.

In looking at the community variables, we witness that urban residence has a
significant negative impact on cumulative fertility in almost all of our regressions and at
all attained ages. Protestant and Catholic background has either no relation or positive
influence on fertility. However, Islamic background has significant negative influence on
fertility in almost all of our regressions. Our results indicate that female schooling alone,
controlling for relative cohort position, can account for about 23 percent to 61 percent of
the decline in fertility.

Similar observations are made for the relationship between female schooling and
age at first birth and age at first marriage. For instance at age 18, 1-3 years of female
schooling either ushers women into their first marriages early or has no predictive power.

At age 23, 1-5 years of female schooling has no predictive influence on first marriages.

176

With regards to first births our study also provided evidence to support the fact that 1-3
years of schooling has no influence on first births at age 18. In contrast, 1-10 years of
schooling has no influence on first births at age 23.

We also observed that among the community variables urban residence made a
difference in the first marriages and first births at age 18 for women ages 18-47 and those
ages 38-47. Islam also made a difference on the timing of first births and first marriages
but none of the other religious backgrounds made a significant impact on the timing of
first births and first marriages at age 18 and 23.

The inclusion of husband’s education in the regressions for the ever-married
showed that for ever-married women at age 18, only women whose husbands had primary
schooling have fewer children. At ages 23 and 28, it is only women whose husbands
have eleven or more years of schooling that have fewer children. However at ages 33 and
38 husband’s schooling does not make any negative impact on fertility. In fact at age 38,
women whose husbands have 1-5 years of schooling have more children. Polygamous
union has no predictive influence on fertility for the ever-married women.

We also investigated if women behaved differently in fertility decision making
depending on their educational level and also wanted to investigate if fertility had fallen
among the better educated women. A look at cumulative fertility at various attained ages
for levels of female schooling revealed that at ages 18 and 23, there is a perfect negative
relationship between female schooling and fertility. However, at ages 28 and 33, the
negative relationship between female schooling and fertility diminishes except when the

woman has seven or more years of schooling.

177

We found from our multivariate analysis that relative to women ages 43 -47,
younger women with either no schooling or 1-6 years of schooling have more children
born at ages 18, 23 and 28. However, consistently we observed that among women with
seven or more years of schooling, the younger women have fewer children relative to the
oldest women. We also observed that among women in each education category, the
higher educated women had fewer children. There is therefore evidence to suggest that
fertility may have fallen at ages 18, 23 and 28 for women with seven or more years of
schooling.

Community variables including urban residence and religious affiliation seem to
have a greater impact on fertility decisions for women with no schooling or 1—6 years of
schooling. Urban residence is important in all'schooling groups but religious influence
had no predictive impact on fertility decisions for women with seven or more years of
schooling. The net cost of children may be relatively lower in the rural areas because
education may bring fewer employment opportunities in rural areas as a result education
may have a greater negative impact on fertility in the urban compared to the rural areas.

Similarly an investigation into the marital patterns of women with no schooling or
1-6 years of schooling revealed that younger women do first marry earlier relative to the
oldest women. On the other hand, for women seven or more years of schooling, the
younger women first marry later than the oldest women. An observation we also made
for first births.

We also found strong evidence in support of the negative correlation between

female schooling and cumulative fertility in both the rural and urban areas. The influence

I78

of schooling on fertility is even greater in the rural areas than the urban areas. We found
out that women with two or more years of schooling in the rural areas have fewer
children at age 18, but 1- 5 years of schooling has no predictive influence on fertility for
women in the urban areas. At age 23, women with ten or more years of schooling in the
rural areas have fewer children than women in the urban areas. An observation we
continue to make at ages 28 and 33. None of the religious influence seems to matter

except Islam in few of the regressions for women in both the rural and urban areas.

8.3 Findings and Conclusions

 

Consistent with other research in the area our result indicate that the first few
years of female schooling may not be related to fertility it may sometimes raise fertility.
It is only women with more than primary education who have fewer number of children
(Caldwell et a1, 1992). We found the strongest influence of female schooling for
cumulative fertility at ages 18, 23 and 28. For example, at age 18 we found that women
with eleven or more years schooling have 0.53 fewer children, at age 23 they have 1.23
and at 28 they have 1.39 fewer children. Whilst the influence of lower levels of
schooling diminish with age, we found women with eleven or more years of schooling
in almost all of our regressions having fewer children. At ages 33 and 38 the influence
of female schooling diminishes greatly, a finding which is also consistent with other
empirical work in the literature. The influence of women with exactly seven years of
schooling is sometimes unpredictable in some of our regressions. Women who continue

one year after primary and then drop out of school may not be any different from those

179

who left at the end of the primary education. Self-selection may be a problem, they may
quit school to go and get married and have children since the labor marketability at the
end of primary may be limited (Thomas, 1996).

We also found that urban residence reduces fertility in almost all of our
regressions. Protestant and Catholic influence on fertility is minimal but Islam has a
strong negative impact on fertility an observation that was contrary to our expectations.
Polygamous maniage does not make any difference in fertility decision making among
the ever-married women at various attained ages. Inclusion of husband’s education does
not make any difference except for women whose husbands have eleven or more years of
schoolingwho have fewer children.

We made similar observations with age at first marriage and age at first birth. The
first few years of female schooling did not influence marital and birth decision. It is
women with more than primary education who delay both first marriages and first births.
A look at whether younger women have fewer children at each attained age also revealed
that for women with either no schooling or 1-6 years of schooling the younger women
have more children than the oldest women. On the other among women with seven or

more years of schooling we observed that fertility has declined with the younger women.

7.4 Implications and Smutions for Further Research

 

 

There are several implications from our present research findings for public
policy. If Nigeria seeks to lower fertility, investment in programs that will enhance

opportunities for young women to be in school at least to secondary school level (high

180

school) is important. To the extent that there is a boy/girl discrimination when it comes
to parents paying for their children’s education, economic hardships may further hinder
school going opportunities for girls in Nigeria and in Africa in general.

Economic hardship or decline that may affect different socio-economic groups
differentially will affect school going differentially for girls. The greatest financial
impact of an economic decline on schooling may be felt at the secondary school level.
According to the educational figures, about 62 percent of school going girls attended
primary but only 17 percent of them attended secondary (1991 figures). This may be in
part because of higher cost associated with secondary education relative to that of
primary. The boarding and lodging cost associated with secondary education may be an
added incentive in the face of an economic decline for parents to withdraw their girls
from schooling. This in the long-run may have an adverse effect on fertility. If girls quit
school to marry, this may lead to an increase in fertility. On the other hand, economic
hardship in addition to high bride price may lead to postponement of marriage by men.
This may in the short-run lead to a decline in fertility, however, in the long-run its effect
on fertility may be negligible.

However, if a sustainable fertility transition is a policy objective, then programs
should be in place to subsidize female schooling at least at the secondary school level
where the drop out rate is highest to ease the direct economic burden on parents.
Subsidies Will be necessary in the face of credit constraints. Students loans may not be
readily available in the Afiican setting for parents to borrow to pay for their children's

education. If even they could borrow most parents may not be willing to do that for the

181

education of their girls. This is because education is seen everywhere as an investment in
children, which in the African situation, it is expected to yield direct returns to parents
when old in the form of remittances. However, studies have shown that girls may not be
reliable to pay back in the future, marriage and childbearing may reduce their
sovereignty and earning ability. The private benefits and social benefits from girls
education may thus be different. There should also be in addition to subsidizing of girls
education, better job opportunities for women when they graduate from college and
expansion of credit facilities for parents to borrow money to pay for their children's
education

It is also worth investigating further the decline in fertility especially for the
younger women in the survey. Our results indicate a decline in fertility among the
younger women. There is the need for a follow up research to examine if the fertility
patterns now for the younger women will remain the same or will change with time. The
next NDHS, which is yet to be fielded will be a good source of information for a follow-
up research. In addition Nigeria has undergone many economic and political changes
since independence in 1960 that may have had both short and long-run impacts on
fertility. For example, macro-economic shocks from the oil bump and subsequent oil
slump, the Nigerian Civil War and many political instabilities. It is important to research
further the links between these macroeconomic shocks and fertility in Nigeria.

To also investigate the possibility of female schooling as a major determinant of a
fertility transition in sub.Saharan, it will be worth applying the same research procedure

to other Afiican countries before a generalization could be made. It is worth researching

182

 

further into programs and incentives that will make girls stay in school beyond the
primary level. To investigate among which programs such as free textbooks, free school
clothes, or subsidized school has the greatest impact towards the enrollment and retention
of girls in school.

A sustainable fertility decline in Nigeria and sub-Saharan Africa in general will
need a balance between better family planning programs and also female education.
International agencies such as the International Planned Parenthood Federation, in their
fight against high population growth rate in Afiica should also aim at channeling some of
the resources into programs that will enhance both the use of family planning methods
and school enrollment and retention for girls. This may take the form of both direct
investment in family planning programs and indirect investments in the form of subsidies
for girls education to ease the economic burden that female education may impose on

parents.

183

APPENDIX

184

TABLE A 5.1
Distribution of Women by Years of Schooling. Female Cohorts Born 1943-72.

 

 

Nigeria, 1990.
Birth Sunple meanyrs. Std. % '/o %
Cohen Size of Sch. Dev. Oyrs <_3yrs _>_6yrs
ALL WOMEN (18-47)
Age Group
43-47 1943-47 572 1.47 3.04 76.1 80.9 16.3
3842 1948-52 950 1.68 3.16 73.0 78.5 18.0
33-37 1953-57 955 2.39 3.84 64.9 70.8 24.2
28-32 1958-62 1584 2.85 4.16 61.4 66.2 30.1
23-27 1963-67 1682 4.68 4.77 44.5 47.4 48.9
18-22 1968-72 1826 5.54 4.54 33.6 36.4 59.1
Total 1943-72 7572 3.59 4.44 53.9 58.1 38.0
LAGOS
Age Group
4347 1943-47 51 6.45 4.47 23.5 25.5 74.5
38-42 I948-52 1 16 5.00 4.25 33.6 37.9 57.8
33-37 1953-57 140 6.16 4.70 25.7 28.6 65.0
28-32 1958-62 246 6.66 4.58 22.8 26.8 69.1
23-27 1963-67 306 8.68 3.83 9.8 10.8 88.6
18-22 1968-72 333 8.88 3.43 9.0 9.3 88.9
Total 1943-72 1192 7.57 4.30 19.0 19.0 78.3

OTHER URBAN

Age Group

4347 1943-47 120 2.00 3.27 65.0 70.8 22.5
38-42 I948-52 212 2.81 3.79 57.6 62.7 30.7
33-37 1953.57 225 3.60 4.25 49.3 56.0 37.3
28-32 1958-62 390 3.96 4.57 49.2 53.3 43.3
23-27 1963-67 399 6.06 4.81 31.6 34.8 62.7
18-22 1968-72 502 6.98 4.31 20.9 22.9 72.7
Total 1943-72 1848 4.93 4.67 39.7 43.6 52.0
RURAL

Age Group ,

43-47 1943-47 401 0.67 1.91 86.0 91.0 7.0
38-42 [948-52 622 0.67 1.87 85.5 91.5 6.3
33-37 1953-57 590 1.03 2.46 80.2 86.4 9.5
28-32 I958-62 948 1.40 2.91 76.5 81.8 14.6
23-27 1963-67 977 2.86 4.00 60.6 64.0 30.9
18-22 1968-72 994 3.69 4.05 48.3 52.2 42.2
Total 1943-72 4532 2.00 3.43 69.4 74.2 21.7

 

185

Table A5.1. continues.

 

 

Birth Sample mean yrs. Std. % % % % %
Cohort Size of Sch. Dev. Oyrs f_3yrs >_6yrs
SOUTHEAST
Age Group
4347 1943-47 169 1.33 2.64 72.8 82.8 13.6
38-42 1948-52 251 1.41 2.61 70.5 82.9 12.8
3342 1953-57 275 2.48 3.46 55.6 68.0 23.6
28-32 1958-62 390 3.47 3.95 46.4 55.6 36.7
23-27 1963-67 426 5.98 4.03 20.4 25.8 63.6
18.2 1968-72 485 6.66 3.75 14.0 19.4 71.3
Total 1943-72 1996 ' 4.20 4.17 39.5 47.9 44.1
OTHER SOUTHWEST
Age Group
4347 1943-47 119 2.38 3.25 58.0 64.7 26.9
38-42 1948-52 174 3.18 3.66 48.3 56.3 34.5
33-37 1953-57 150 3.71 4.18 45.5 54.0 38.0
2832 1958-62 198 . _ 5.16 4.73 34.3 41.9 55.1
23-27 1963-67 212 7.84 4.24 15.6 18.4 80.7
1 8-22 1968-72 276 8.51 3.12 4.4 6.2 90.4
Total 1943-72 1129 . 5.69 4.54 29.6 35.0 60.2
NORTH
Age Group
43-47 1943-47 233 0.86 0.09 99.1 100.0 0.0
38-42 1948-52 409 026 1.44 96.1 96.8 2.9
33-37 1953-57 390 0.45 1.91 93.1 94.4 4.6
28-32" 1958-62 750 0.66 2.12 89.1 91.1 7.3
23-27 1963-67 738 1.35 3.10 81.0 83.3 14.9
18-22 1968-72 735 2.16 3.52 68.7 71.2 25.4
Total ' 1943-72 3255 1.03 2.69 84.7 86.6 11.7

 

Source : Calculated from Demographic and Health Survey, Nigeria, 1 990.

186

TABLE A 5.2

Percentage of All Women with Second and Third or more Births at Various

 

 

 

 

 

 

 

Attained Ages
SECOND BIRTHS THIRD OR MORE BIRTHS
BY AGE BY AGE
182328 33 18232833 N

ALL SAMPLE
Age group
18-22 10.5 - - - 2.3 - - - 1831
23-27 14.0 52.1 - - 4.0 29.0 - - 1684
28-32 15.8 57.9 81.3 - 4.3 36.4 67.0 - 1587
33-37 14.0 54.0 82.2 92.5 2.9 33.1 70.5 85.2 955
38-42 14.0 47.7 73.9 87.6 4.2 28.9 59.7 78.2 951
43-47 12.2 47.2 71.3 83.6 3.8 26.9 55.9 74.5 572
SOUTHEAST
Age Group
18-22 7.6 - - - 1.6 - - - 487
23-27 10.5 49.4 - -- 3.0 28.1 - -- 427
28-32 14.8 57.9 82.9 - 5.4 39.0 68.4 - 392
33-37 14.5 55.3 83.6 92.0 4.4 36.7 72.0 84.7 275
38-42 ' f 21.1'57.4 83.3 93.6 7.6 37.8 72.1 90.8 251
43-47 16.0 53.8 80.5 91.1. 6.5 34.9 65.7 85.2 169
SOUTHWEST
Age Group
18-22 . 1.1 - - - 0.2 - -- - 609
23-27 4.8 32.9 - - 0.4 12.1 - - 519
28-32 7.0 46.3 77.8 - 1.1 22.5 61.3 - 445
33-37 7.6 51085.2 96.6 0.7 25.9 72.4 91.0 290
38-42 4.1 39.9 78.0 93.8 0.7 17.2 58.1 81.8 291
43-47 4.7 49.4. 77.6 91.2 0.6 20.6 57.6 80,0 170
NORTH
Age Group
18-22 20.1 - -- -- 4.6 - --- -- 735
23-27 22.4 67.2 - -- 7.2 41.3 - -- 738
28—32 21.5 64.8 82.7 -- 5.6 43.3 69.7 -- 750
33-37 18.5 55.4 79.0 89.7 3.6 35.9 67.9 81.3 390
38-42 16.6 47.4 65.3 79.5 4.6 31.8 53.3 68.0 409
43-47 15.0 40.8 60.1 72.5 4.3 25.8 47.6 62.7 233

 

187

Predictions of the Decline in Fertility Attributable to Female Schooling:

TABLE A6.l

Cumulative Fertility at Age 18

 

 

 

 

Female Schooling In Regression Means of Estimated Means of Estimated Change In
Single Years Coefficients Female Means Female Means Fertility
Women Ages Schooling Col 2*col 3 Schooling Col 2* Col 5 Col 4 -
18-47 Women Women Ages Col 6
ages 18-22 38-47
1 year of Schooling 0.358 0.002 0.0007 0.012 0.0043 -0.0036
2 years of Schooling -0.106 0.009 -0.0010 0.028 -0.0030 0.0020
3 years of Schooling -0.092 0.018 -0.0017 0.045 -0.0041 0.0025
4 years of Schooling -0.11 0.031 -0.0034 0.033 -0.0036 0.0002
5years of Schooling -0.276 0.023 -0.0064 0.022 -0.0061 -0.0003
6 years of Schooling -0.248 0.217 -0.0538 0.178 -0.0441 -0.0097
7 years of Schooling -0.222 0.027 -0.0060 0.009 -0.0020 -0.0040
8 years of Schooling -0.469 0.055 -0.0258 0.010 -0.0047 -0.0211
9 years of Schooling -0.359 0.075 -0.0269 0.022 -0.0079 -0.0190
10 years of Schooling -0.526 0.065 -0.0342 0.008 -0.0042 -0.0300
11+ years of -0.529 0.377 -0.1994 0.059 -0.0312 -0.1682
Schooling
Sum -0.3578 -0.1067 -0.2511
Female’ Schooling In Regression Means of Estimated Means of Estimated Change In
Single Years Coefficients Female Means Female Means Fertility
Women Ages Schooling C012‘c013 Schooling Col 2* C015 Col 4 -
18-22 Women Women Col 6
ages 18-22 Ages
38-47
1 year of Schooling ‘ 0.464 0.002 0.0009 0.012 0.0056 -0.0046
2 years of Schooling ' -0.236 0.009 -0.0021 0.028 -0.0066 0.0045
3 years of Schooling 0.064 0.018 0.0012 0.045 0.0029 -0.0017
4 years of Schooling -0.183 0.031 -0.0057 0.033 -0.0060 0.0004
5years of Schooling ' -0.138 0.023 -0.0032 0.022 -0.0030 -0.0001
6 years of Schooling -0.213 0.217 -0.0462 0.178 -0.0379 -0.0083
7 years of Schooling -0.236 0.027 -0.0064 0.009 -0.0021 -0.0043
8 years of Schooling -0.403 0.055 -0.0222 0.01 -0.0040 -0.0181
9 years of Schooling -0.317 0.075 -0.0238 0.022 -0.0070 -0.0168
10 years of Schooling -0.466 0.065 -0.0303 0.008 -0.0037 -0.0266
11+ years of -0.495 0.377 -0.1866 0.059 -0.0292 -0.1574
Schooling
Sum -0.3243 -0.0912 -0.2331

 

I88

Predictions of the Decline in Fertility Attributable to Female Fertility:
Cumulative Fertility at Age 23

TABLE A62

 

 

 

 

Female Schooling 1n Regression Means of Estimated Means of Estimeated Changs
Single Years Coefficients Female Means Female Means In
Women Schooling Col 2" Col 3 Schooling Col 2 ' Col Fertility
Ages Women Women 5 Col 4 -
23-47 Ages 23-27 Ages 38-47 Col 6
1 year of Schooling 0.618 0.001 0.0006. 0.012 0.0074 -0.0068
2 years of Schooling 0.227 0.012 0.0027 0.028 0.0064 -0.0036
3 years of Schooling 0.028 0.021 0.0006 0.045 0.0013 -0.0007
4 years of Schooling 0.069 0.029 0.0020 0.033 0.0023 -0.0003
5 years of Schooling -0. 163 0.023 -0.0038 0.022 -0.0036 -0.0002
6 years of Schooling -0.238 0.249 -0.0593 0.178 -0.0424 -0.0169
7 years of Schooling -0.187 0.005 -0.0009 0.009 -0.0017 0.0007
8 years of Schooling -0.544 0.033 -0.0180 0.01 -0.0054 -0.0125
9 years of Schooling -0.477 0.072 -0.0343 0.022 -0.0105 -0.0239
10 years of Schooling -0.721 0.038 -0.0274 0.008 -0.0058 -0.0216
11+ years of Schooling -1.257 0.357 -0.4488 0.059 -0.0742 -0.3746
, Sum -0.5865 -0.1262 -0.4603
Female Schooling In Regression Means of Estimated Means of Estimeated Changs
Single Years Coefficients Female Means Female Means In
Women Schooling Col 2* Col 3 Schooling Col 2 * Col Fertility
Ages Women Women 5 Col 4 -
23-27 Ages 23-27 Ages 38-47 Col 6
1 year of Schooling -0.218 0.001 -0.0002 0.012 -0.0026 0.0024
2 years of Schooling 0.450 0.012 0.0054 0.028 0.0126 -0.0072
3 years of Schooling 0.063 0.021 0.0013 0.045 0.0028 -0.0015
4 years of Schooling 0.036 0.029 0.0010 0.033 0.0012 -0.0001
5 years of Schooling -0.368 0.023 -0.0085 0.022 -0.0081 -0.0004
6 years of Schooling -0.610 0.249 -0.1519 0.178 -0.1086 -0.0433
7 years of Schooling -1.187 0.005 -0.0059 0.009 -0.0107 0.0047
8 years ofSchooling -0.703 0.033 -0.0232 0.010 -0.0070 -0.0162
9 years of Schooling ' ' -0.790 0.072 -0.0569 0.022 -0.0174 -o.0395
10 years of Schooling -0.853 0.038 -0.0324 0.008 -0.0068 -0.0256
11+ years of Schooling -1.620 0.357 -0.5783 0.059 -0.0956 -0.4828
Sum -0.8496 -0.2402 -0.6094

 

I89

TABLE A6.3
Impact of Education on Fertility - Female Schooling, cohort effects and
community variables. (RURAL RESIDENCE)

 

 

 

 

Dependent Variable Cumulative Cumulative Cumulative Cumulativ Cumulative
Fertility At Fertility at Fertility at e Fertility Fertility at
Age 18 Age 23 Age 28 at Age 33 Age 38
18 - 47 23-47 28-47 33-47 38-47
IndeEgdent Variables:
Dummies for:
1 year ofSchooling 0.357‘“ 0.639‘” 1.025” 1.865” 2.157“
(0.19) (0.36) (0.49) (0.66) (0.95)
2 years of Schooling - 0.210‘“ 0.029 0.132 0.001 0.266
(0.11) (0.21) (0.30) (0.38) (0.54)
3 years of Schooling - 0.164‘“ - 0.002 0.075 0.970" 1.449"
(0.09) (0.19) (0.27) (0.38) (0.52)
4 years of Schooling - 0.166'” - 0.131 - 0.020 0.025 - 0.561
(0.10) (0.20) (0.33) (0.46) (0.73)
5 years of Schooling - 0.415‘ - 0.417‘” - 0.306 0.387 0.306
(0.1 1) (0.23) (0.38) (0.54) (0.80)
6 years of Schooling - 0.331‘ - 0.451' - 0.190 0.385 0.608
(0.06) (0.11) (0.18) (0.27) (0.39)
7 years of Schooling 0.002 0.955 1.823 6.882‘ 5.804”
(0.20) (0.68) (I . 16) (2.26) (2.49)
8 years of Schooling - 0.409“ - 0.591 - 0.494 - 4.196‘” - 4.532‘”
(0.15) (0.39) (0.90) (2.26) (2.49)
9 years of Schooling - 0.362‘ - 0.214 - 1.156“ - 0.267 1.325
‘ (0.11) (0.26) (0.54) (0.75) (1.12)
10 years of Schooling - 0.592‘ - 0.897“ - 0.824 0.802 - 1.019
(0.14) (0.34) (0.90) (2.00) (2.49)
11 or more years - 0.605‘ - 1.682“ - 2.014' . 1.968” - 1.610‘
of Schooling (0.08) (0.17) (0.36) (0.72) ( I .25)
Cohort Effects
18-22 0.078
(0.08)
23-27 0139‘“ 0.145
(0.08 (0.14)
28-32 0.194“ 0.340“ 0.325‘“
(0.07) (0.13) (0.18)
33-37 0.041 - 0.025 0.264 0.082
(0.08) (0.14) (0.18) (0.21)
38-42 0.120 - 0.034 0.035 - 0.001 - 0.176
(0.07) (0.14) (0.18) (0.20) (0.23)
Community Variables
Protestant - 0.025 0.163 0.268 0.139 - 0.109
(0.06) (0.12) (0.18) (0.25) (0.34)
Catholic 0.016 0.200 0.212 0.061 - 0.446
(0.07) (0.13) (0.20) (0.26) (0.38)
Islam - 0.221“ - 0.224 - 0.151‘ - 0.316 - 0.528
(0.08) (0.16) (0.24) (0.31) (0.42)
intercept 0.619‘ 1.888' 2.377‘ 5.057‘ 6.640‘
(0.08) (0.14) (0.20) (0.25) (0.32)
R-squared 0.07 0.08 0.04 0.05 0.06
F( all Covariates) 7.44 7.71 3.01 2.60 1.98
N 1966 1530 l 140 773 492

 

Note : Standard errors in parenthesis. '-p-va1ue < 0.01, “- p-value < 0.05,

190

”‘-— p-vaiue < 0.10.

TABLE A6.4
Impact of Education on Fertility - Female Schooling, cohort effects and
community variables. (URBAN RESIDENCE)

 

 

 

 

 

 

Dependent Cumulative Cumulative Cumulative Cumulative Cumulative
Variable Fertility At Fertility At Fertility At Fertility At Fertility At
Age 18 Age 23 Age 28 Age 33 Age 38

18 - 47 23-47 28-47 33-47 38-47

Independent Variables:

1 year of Schooling 0.218 0.332 0.621 0.678 1.230
(0.24) (0.52) (0.72) (0.92) (1.18)

2 years of Schooling 0.099 0.519 0.757 0.997 - 0.313
(0.14) (0.33) (0.50) (0.84) (1.65)

3 years of Schooling 0.002 - 0.018 0.129 0.31 1 0.216
(0.09) (0.19) (0.29) (0.40) (0.63)

4 years of Schooling - 0.026 0.295 0.535 1.047 1.734"
(0.08) (0.20) (0.30) (0.40) (0.60)

5 years of Schooling - 0.093 0.057 0.131 - 0.012 - 0.458
(0.10) (0.24) (0.3 5) (0.50) (0.80)

6 years of Schooling - 0.109‘ - 0.004 0.081 0.171 0.175
(0.04) (0.09) (0.14) (0.50) (0.30)

7 years of Schooling - 0.217“ - 0.322 - 0.006 0.520 0.795
(0.10) (0.30) (0.45) (0.66) (0.90)

8 years of Schooling - 0.378‘ - 0.372” - 0.164 - 0.093 0.613
(0.07) (0.19) (0.33) (0.50) (0.84)

9 years of Schooling - 0.243‘ - 0.389“ - 0.096 - 0.236 - 0.711
(0.06) 0. 15) (0.26) (0.41) (0.65)

10 years of Schooling - 0.397‘ - 0.523“ - 0.683 - 0.740 0.242
(0.07) (0.20) (0.37) (0.54) (0.96)

11 or more years - 0.389‘ - 0.984‘ - 1.168‘ - 1.036‘ - 0.978“

of Schooling (0.04) (0.09) (0.16) (0.25) (0.39)

Cohort Effects

18-22 - 0.009
(0.06)

23-27 0.056 0.061
(0.06) (0.13)

28-32 0.076 0.162 0.475 “

' (0.06) (0.13) (0.18)

33-37 0.1 11‘“ 0.348 0.864' 0.955
(0.06) (0.13) (0.19) (0.21)

38-42 - 0.021 - 0.060 0.283 0.528 0.311
(0.06) (0.14) (0.19) (0.22) (0.25)

Community Variables

Protestant 0.073 0.098 0.065 - 0.519 - 0.428
(0.09) (0.23) (0.37) (0.50) (0.73)

Catholic 0.089 0.300 0.381 - 0.026 0.037
(0.10) (0.23) (0.38) (0.52) (0.77)

Islam 0.029 0.049 - 0.180 - 1.027” - 1.193”
(0.10) (0.23) (0.37) (0.50) (0.74)

Intercept 0.360‘ 1.477‘ 2.739‘ 4.548” 5.695'
(0.10) (0.24) (0.38) (0.50) (0.71)

R-squared 0.09 0.12 0.10 0.10 0.09

F( all Covariates) 1 1.46 I 1.29 7.48 4.67 2.42

N 2351 554 1 139 672 388

 

Note: Standard errors in parenthesis. ‘-p -va1ue < 0.01, “- p-vaiue < 0.05, “‘-— p-value < 0.10..

191

TABLE A 6.5

Impact of Education on Fertility: Female Schooling Result by Age Cohort

 

 

 

 

Dependent Variable CumulativeFertiliy at Age 18 Cumulative Fertility Cumulative
at Age 23 Fertility at
Age
28
18-22 23-27 28-32 23-27 28-32 28-32
Indgpendent Variables:
1 year of Schooling 0.464 - 0.720 - 0.378 - 2187’” - 0.517 - 0.457
(0.37) (0.66) (0.32) (1.25) (0.58) (0.75)
2 years of Schooling - 0.236 - 0.098 - 0.045 0.450 0.261 0.654
(0.17) (0.20) (0.19) (0.39) (0.34) (0.44)
3 years of Schooling 0.064 - 0.120 - 0.225 0.063 - 0.319 - 0.385
(0.13) (0.16) (0.14)‘” (0.30) (0.24) (0.32)
4 years of Schooling - 0.183’” 0.132 - 0.211 0.036 - 0.017 0.335
(0.10) (0.14) (0.17) (0.26) (0.29) (0.39)
5 years of Schooling - 0.136 - 0.356” - 0.348“ - 0.368 - 0.047 0.164
(0.1 1) (0.15) (0.17) (0.28) (0.31) (0.40)
6 years of Schooling - 0.213‘ - 0.373‘ - 0.306’ - 0.610‘ - 0.435’ - 0.400“
(0.06) (0.07) (0.07) (0.13) (0.13) (0.17)
7 years of Schooling - 0.236” - 0.720” - 0.087 - 1.187” - 0.559 - 0.915
(0.1 1) (0.30) (0.28) (0.57) (0.50) (0.65)
8 years of Schooling - 0.403‘ - 0.494‘ - 0.600‘ - 0.703‘ - 0.628‘" - 0.512
' (0.08) (0.13) (0.19) (0.25) (0.34) (0.44)
9 years of Schooling - 0.317‘ - 0.470‘ - 0.493‘l - 0.790" - 0.621” - 0.853”
(0.08) (0.10) (0.15) (0.18) (0.26) (0.34)
10 years of Schooling - 0.466‘ - 0.634' - 0.635‘ - 0.853‘ - 1.107” - 1.405”
(0.08) (0.12) (0.22) (0.23) (0.40) (0.52)
11 or more years of - 0.495‘ - 0.658' - 0.583‘ - 1.620‘ - 1.440‘ - 1.873‘
Schooling (0.06) (0.06) (0.08) (0.12) (0.14) (0.18)
Intercept 0.536‘ 0.720‘ 0.71 1‘ 2.187‘ 2.184‘ 3.790‘
(0.05) (0.05) (0.045) (0.10) (0.08) (0.10)
R-squared 0.10 0.13 0.08 0.21 0.13 0.14
F( all Covariates) I 1.21 13.07 6.65 22.69 1 1.05 12.08
N 944 834 944 834 834

1094

 

Note: Standard Errors in parenthesis. ‘- p-vaiue < 0.01, " - p-value < 0.05, "‘ - p-value < 0.10..

192

TABLE A6.6
Impact of Education on Fertility - Female Schooling
and Cohort Effect : Results by Age Cohort

 

 

 

 

Dependent Variable Cumulative Fertility At Age 18 Cumulative Fertility At Age 23
18-27 28-37 3847*- 23-27 28-37 38-47
Independent Variables:
1 year of Schooling 0.039 0.124 0.709" - 2.187‘” . 0.010 1.613‘
(0.34) (0.23) (0.26) (1.25) (0.42) (0.45)
2 years of Schooling - 0.068 - 0.01s - 0.277 0.450 0.324 - 0.044
(0.13) (0.13) (0.17) (0.39) (0.246) (0.31)
3 years of Schooling - 0.038 - 0.130 - 0.094 0.063 0.001 0.01 1
(0.10) (0.10) (0.14) (0.30) (0.19) (0.25)
4 years of Schooling - 0.061 — 0.056 - 0.360” 0.036 0.267 - 0.400
(0.08) (0.1 l) (0. l 6) (0.26) (0.21) (0.29)
5 years of Schooling - 0.253” - 0.244” - 0.455“ - 0.368 0.037 - 0.545
(0.09) (0.13) (0.20) (0.28) (0.24) (0.35)
6 years of Schooling - 0.302‘ - 0209‘ - 0.237‘ - 0.610‘ - 0.198‘” - 0.059
(0.05) (0.06) (0.08) (0.13) (0.10) (0.51)
7 years of Schooling - 0359' - 0.182 0.182 - 1.187" . 0.297 0.509
(0.1 1) (0.24) (0.30) (0.57) (0.44) (0.53)
8 years of Schooling - 0.463’ - 0.560‘ - 0560“ - 0.703‘ - 0.555“ - 0.757
(0.07) (0.15) (0.28) (0.25) (0.28) (0.50)
9 years of Schooling - 0.402‘ - 0.449’ - 0.039 - 0.790‘ - 0.530“ - 0.062
- (0.06) (0.12) (0.20) (0.18) (0.213) (0.35)
10 years of Schooling - 0.553‘ - 542‘ - 0.567'" - 0.853‘ - 0.840” - 0.887
(0.07) (0.17) (0.32) (0.23) (0.304) (0.57)
1 1 or more years of - 0.584‘ - 0.470‘ - 0.449‘ - 1.620’ - 1.199‘ - 0.728‘
Schooling (0.04) (0.06) (0.12) (0.12) (0.12) (0.22)
Cohort Effects
18-22 - 0.060”
' (0.05)
23-27
28-32 0045 0.064
(0.04) (0.08) -
33-37
38-42 0.043 - 0.085
(0.06) (0.10)
Intercept 0.668‘ 0.649‘ 0.536‘ 2.187‘ 1.978‘ 1.805‘
(0.04) (0.04) (0.05) (0.10) (0.07) (0.09)
R-squared 0.12 0.06 0.05 021 0.09 0.04
F( all Covariates) 22.32 6.78 3.69 22.69 1 1.23 2.89
N ‘ 2037 l 399 880 944 1399 880

 

Note: Standard Errors in parenthesis. ’— Coefficient is significmt at 0.01, “— Coefficient is significant at 0.05. “-
Coeflicient is significant at 0.10.

193

TABLE A6.7
Impact of Education on Fertility - Female Schooling, Cohort Eeffects
and Community Variables: Results by Age Cohort

 

 

 

 

 

Cumulative Fertility Cumulative Fertility Cumulative Fertility
At Age 18 At Age 23 At Age 28
28 - 37 38-47 28-37 38-47 28-37 3847
Independent Variables:
1 year of Schooling 0.098 0.676“ - 0.065 1.551‘ 0.191 1.701 ‘
(0.23) (0.25) (0.42) (0.45) (0.53) (0.60)
2 years of Schooling - 0.032 - 0.356" 0.273 - 0.203 0.520 - 0.058
(0.13) (0.17) (0.25) (0.31) (0.32) (0.36)
3 years of Schooling - 0.128 - 0.1 17 - 0.017 - 0.031 - 0.023 0.287
(0.11) (0.14) (0.18) (0.24) (0.24) (0.32)
4 years of Schooling - 0.059 - 0.323" 0.243 - 0.324 0.430 - 0.122
(0.1 I) (0.16) (0.21) (0.28) (0.27) (0.37)
5 years of Schooling - 0.255 - 0.457” - 0.006 - 0.543 0.033 - 0.454
(0.13) (0.19) (0.24) (0.34) (0.30) (0.45)
6 years of Schooling - 0.189' - 0.175“ - 0.198‘” 0.066 - 0.195 0.153
(0.06) (0.08) (0.1 1) ' (0.14) (0.13) (0.19)
7 years of Schooling - 0.151 0.251 - 0.271 0.655 - 0.602 1.343‘”
(0.24) 0.29) (0.44) (0.53) (0.56) (0.70)
8 years of Schooling - 0.523‘ - 0.463‘” - 0.535‘” - 0.561 - 0.408 - 0.168
(0.15) (0.31) (0.28) (0.50) (0.36) (0.66)
9 years of Schooling - 0. 400‘ 0. 055 - 0.483“ 0.119 - 0.748“ 0.335
(0.12) (0.19) (0.22) (0.35) (0.28) (0.46)
10 years of Schooling - 0.495‘ - 0.502 - 0.788" - 0.748 - 1.080‘ - 0.036
. . (0.17) (0.31) (0.31) (0.57) (0.39) (0.74)
11 or more years of. . - 0.432‘ - 0.362" - 1.171‘ - 0.546“ - 1.628‘ - 0.654”
Schooling (0.07) (0.13) (0.13) (0.23) (0.16) (0.30)
Cohort Effects
18-22
23-27
28-32 0.050 0.072 - 0.153
(0.04) (0.08) (0.10)
33-37
38-42 0.064 - 0.046 0.124
(0.06) (0.13)
Community Variables
Urban Residence - 0.087‘” - 0.172“ - 0.144 - 0.364‘ - 0.142 - 0.521‘
(0.05) (0.07) (0.09) (0.12) (0.12) (0.15)
Protestant 0.01 1 0.006 0.226 0.081 0.453” - 0.202
(0.08) (0.10) (0. I 5) (0.18) (0.20) (0.24)
Catholic 0.042 0. 137 0.227 0.301 0.465” 0.007
(0.09) (0. 1 I) (0. 1 7) (0.20) (0.21) (0.26)
Islam - 0.075 - 0.181 - 0.063 - 0.243 0.220 - 0.706”
(0.12) (0.1 1) (0.17) (0.20) (0.22) (0.26)
Intercept 0.636‘ 0.586‘ 1.873‘ 1.851 t 3.529‘ 3.608‘
(0.08) (0.10) (0.13) (0.17) (0.19) (0.22)
R-squared 0.06 0.07 0.09 0.07 0.1 1 0.07
F( all Covariates) 5.70 4.34 8.96 3.90 11.07 3.80
F(Cohort and Community) 2.1 1 4.95 1.78 5.48 2.59 6.70
F(Community) 2.36 6.06 2.06 6.68 2.57 8.31
N 1399 880 1399 880 1399 880

 

Note: Standard Errors in parenthesis. ‘- p-value < 0.01, "- p-value < 0.05, "t“ p-value < 0.10.

194

REFERENCES

Abu Bakar, Noor Laily bt.D, Tan Boon Ann, Tey Nai Peng and Rohani Abd Razak.
1985. "Ethnicity and Fertility in Malaysia." In Ethnicity and Fertility in Southeast Asia
Series. Research Notes and Discussion Paper No. 52. Institute of Southeast Asian
Studies.

Agarwaia, S. 1960. "A Method for Correcting Reported Ages and Marriage Durations."
Indian Population Bulletin 1 (I): 129-164.

Adewuji, Alfred and Bamikaie J. Feyisetan. 1988. "Fertility Differentials Among the
three Major Nigerian Ethnic Groups Resident in Lagos." DSS Monograph series No. 4.
Published by the Department of Demography and Social Statistics, Obafemi Awolowo
University, lie-Ife, Nigeria.

Ainsworth, Martha. 1989. "Socioeconomic Determinants of Fertility in Cote d'Ivoire."
Living Standards Measurement Study (LSMS) Working Paper No. 53. Policy Research
Department. World Bank, Washington, DC.

Ainsworth, Martha, Kathleen Beegle and Andrew Nyamete. 1996. "The Impact of
Female Schooling 0n Fertility and Contraceptive Use: A Study of Fourteen Sub-Saharan
Countries." The World Bank Economic Review (January ) 10 (1): 85-122.

Andoh, Emmanuel Kenneth, 1990. "Response Variability in African Demographic
Survey Data: A Case Study of a Nigerian Village," African Demography Program
Working Paper No. 1. Population Studies Center, University of Pennsylvania.

Arnold, Fred. 1990. "Assessment of the Quality of Birth History Data in the
Demographic and Health Surveys." In An Assessment of the DHS-I Data Quality, 83-
109. IRD/Macro Systems. DHS Methodological Reports No.1 Columbia.

Arowolo, Oladele O. 1976. "Determinants of Fertility Arnong Yorubas of Nigeria," in
Recent Empirical Findings on Fertility: Korea, Nigeria, Tunisia, Venezuela, Philippines.
Occasional Monograph Series No. 7. Interdisciplinary Communications Program.
Smithsonian Institute.

Atado, Rev. Fr. Joe Chuks. 1988. "Afiican Marriage Customs & Church Law."
Published by The Modern Printers Limited. Kano.

Becker, Gary S. 1981. "A Treatise on the Family." Cambridge. Mass: Harvard
' University Press.

. 1965. "A Theory of the Allocation of Time". Economic Journal 75, 493-

 

517.

I95

. 1960. "An Economic Analysis of F ertility," in Demographic and
Economic Change in Developed Countries. Universities National Bureau of Economic
Research Conference Series 11. Princeton, NJ. : NBER : 209-31.

 

Becker, GS and H. Gregg Lewis, 1973. "On the Interaction between Quantity and
Quality of Children," Journal of Political Economy (Mar/April) 81 (2): 8279-5288.

Becker, GS and Nigel Tomes, 1976. "Child Endowments and the Quantity and Quality
of Children," Journal of Political Economy (August) 84(2): 8143-8162.

Beegle, Kathleen. 1994. "The Quality and Availability of F arnily Planning Services and
Contraceptive use in Tanzania." LSMS Working Paper No. 114. Policy Research
Department and Eastern Africa Department, World Bank, Washington, DC.

Behrman, Jere and Anil Deolalikar. 1988. " Nutrition and Health." In Handbook of
Development Economics, Hollis Chenery and TN. Srinivasan (eds) vol 1 :631-711

Benefo, Kofi and T. Paul Schultz. 1996. "Determinants of Fertility and Child Mortality
in Cote d'Ivoire." World Bank Economic Review (January) 10 (1) : 123-158.

Bevan, David, Paul Collier and Jan Willem Gunning. 1992. "Nigeria 1970-1990."
International Center for Economic Growth. Country Studies No. 1 1. ICS Press.

Birdsall, N., and F. T. Sai. 1988. "Family Planning Services in Sub-Saharan Afiica."
Finance and Development (March): 28-31.

' Bongaarts, J. 1994. "Population Policy Options in the Developing World." Science 263,
no. 5148:771-776.

. 1987. "The Proximate Determinants of Exceptionally High Fertility."
Population and Development Review l3(1):133-397.

 

. 1984. "Implications of Future Fertility Trends For Contraceptive Practice."
Population and Development Review 10(2): 341-52.

 

. 1980. "Does malnutrition affect fecundity?" A summary of the evidence."
Science 208:564-569.

 

. 1978. "A Framework for Analyzing the Proximate Determinants of
Fertility." Population and Development Review 4:105-132.

 

Bongaarts, John, Odile Frank and Ron Lesthaeghe, 1984. "The Proximate Determinants
of Fertility in Sub-Saharan Africa". Population and Development Review 10(3):511-537

196

Bongaarts , John and Robert G. Potter (eds). 1983. Fertility, Biology and Behavior: An
Analysis of the Proximate Determinants. New York Academic Press.

Boserup, E. 1985. "Economic and Demographic Interrelations in Sub-Saharan Africa",
Population and Deve10pment Review, 11(3 ), pp. 383-3 97.

. 1970. Women's Role in Economic Development, New York: St. Martin's

 

Press.

Bulatao, R. A, and R. D. Lee (eds). 1983. Determinants of Fertility in Developing
Countries (2 vols). New York : Academic Press.

Busia, EA. 1954. "Some aspects of the relation of social conditions to human fertility
in the Gold Coast." In F. Lorimer (ed) Culture and Human Fertility. UNESCO, Paris,
341-350.

Burch, Thomas K. 1966. "The Fertility of North American Catholics: A Comparative
Overview." Demography 3

Cain, M.’ . 1984. "Women's Status and Fertility in Developing Countries: Sons and

Economic Security." New York: The Population Council. Center For Policy Studies
Working Paper #10.

- . 1983. "Fertility as an Adjustment to Risk." Population and Development
Review 9(4): 688-702.

 

. Caldwell, John C. 1995. "Population Factor in African Change." In Economic and
Demographic Change in Africa. Archie Mafeje and Samir Radwan (eds.) pp.l 1-35.

. 1986, "Routes to Low Mortality in Poor Countries." Population and
Development Review, 12 (2), pp. 171-220.

 

.' 1982. "A Theory of Fertility Decline." Canberra: Australian National
University Press.

 

. . 1980. "Mass Education as a Determinant of The Timing of Fertility
Decline." Population and Development Review 6(2): 225-256.

 

. 1968: "The Control of Family Size in Tropical Africa." Demography 5(2):

 

598-619.

. 1966. "Study of Age Misstatement among Children in Ghana."
Demography (1): 447-490.

 

197

Caldwell, John C. and Pat Caldwell. 1990. "High Fertility in Sub-Saharan Africa."
Scientific American May

. 1987. "The Cultural Context of High Fertility in Sub—Saharan Africa. "
Population and Deve10pment Review 13(3):409-437.

 

. 1977. "The Role of Marital Sexual Abstinence in Determining Fertility: A
Study of the Yoruba in Nigeria." Population Studies, 31 (2), pp 193-217

 

Caldwell, J .C and A.A Igun. 1971. "An Experiment with Census Type Age Enumeration
in Nigeria," Population Studies, vol 25 (2): 297-302.

Caldwell John C. and H. Ware. 1977. "The Evolution of Family Planning in an African
City, Ibadan, Nigeria", Population Studies, 31(3), pp. 487-507.

Caldwell, John. C., 1.0. Orubuloye, and Pat. Caldwell. 1991. "Fertility Decline in Africa:
A New Type of Transition", Population and Development Review, 18(12), pp. 211-42

Casterline, J. B. 1985. "Community Effects on Fertility." In The Collection and Analysis
of Community Data. Casterline, J .(ed). Voorburg, Netherlands: International Statistical
Institute.

Chou, Ru-Chi, and Susannah Brown. 1968. "A Comparison of the Size of Families of
Roman Catholic and Non-Catholics in Great Britain." Population Studies, 22:51-60.

Cleland, J. 1985. "Marital Fertility Decline in Developing Countries: Theories and
Evidence." pp. 223-252. In Cleland, J. and Hobcraft, J. (eds). Reproductive Change in
Developing Countries. Insights From the World Fertility Survey. New York: Oxford
University Press.

Cleland, J. G and V.C. Chidambaram. 1981. " The Contribution of the World Fertility
Surveys to an Understanding of Fertility Determinants and Trends." Paper presented at
the IUSSP General Conference, Manila.

Cleland, J. and J. K. van Ginneken. 1988. "Maternal Education and Child Survival in
Developing Countries: The Search for Pathways of Influence." Social Science and
Medicine. 27 (12). pp.1357-68. *

Coale, Ansley J ., 1986. "The Population trends and economic development." In Jane
Menken (ed), World Population and US. Policy: The Choice Ahead. New York: W.W.
Norton, pp. 96-104. .

Cochrane, Susan Hill 1979. "Fertility and Education: What do we Really Know?" World
Bank Staff Occasional Papers No. 26. Baltimore: Johns Hopkins University Press.

198

Cochrane, Susan H. and Samir Farid. 1990a. "Socioeconomic Determinants in Fertility
and their Explanation." In Ascadi et al. (ed), pp.144-154

. 1990b. "Fertility in Sub-Saharan Africa: Analysis and Explanation."
Washington DC: World Bank Discussion Papers No. 43.

 

Cramer, J. C. 1980. "Fertility and Female Employment: Problems of Causal Direction."
American Sociological Review 45 (April): 167-190.

Das, N. 1987. "Sex Preference and Fertility Behavior: A Study of Recent Indian Data."
Demography 24 (4): 518-30.

Davis, K. 195 5. "Institutional Factors Favoring High Fertility in Underdeveloped Areas."
Eugenics Quarterly2: 33-39.

Davis, K. and J. Blake. 1956. "Social Structure and Fertility." Economic Development
and Cultural Change 4:211-235.

Day, Lincoln H. 1964. "Fertility Differential Among Catholics in Australia." Milbank
Memorial Fund Quarterly (April) 42 (2): 57-63.

Europa Publications. 1996. Africa South of the Sahara 1997. Twenty-Sixth Edition
Europa Publications Limited, 18 Bedford Square, London, England.

Easterlin, Richard A. 1980. Birth and Fortune: The Impact of Numbers on Welfare. New
York: Basic Books.

. 1968. "Population, labor force, and long savings in economic growth. The
American Experience." New York National Bureau of Economic Research.

 

. 1967-.- "Effects of Population Growth on the Economic Development of
‘ Developing Countries." Annals of the American Academy of Political and Social Science
369: 98-108. -

 

Ebigbola, J.A. and A.K. Omideyi. 1988. "Fertility Behavior in an Urban Center: A Case
' Study of Married Women in Ilesa." DSS Monograph No. 3. Obafemi Awolowo
University, lie-Ife, Nigeria.

Ewbank, DC 19981. "Age Misreporting and Age Selective Underenumeration:
Sources, Patterns and Consequences for Demographic Analysis." Committee on

Population and Demography, Report No. 4 Washington DC National Academy Press.

Farooq, Ghazi D. 1985. "H0usehold Fertility Decision-Making in Nigeria." In Farooq
and Simmons, eds., Fertility in Developing Countries, London: MacMillan Press.

199

Federal Office of Statistics of Nigeria. 1992. Nigeria Demographic and Health Survey.
IRD/Macro International Inc. NDHS 1990. Lagos and Columbia, MD.

Feyisetan, Bamikaie and Martha Ainsworth. 1996. "Contraceptive Use and the Quality,
Price, and Availability of Family Planning in Nigeria." World Bank Economic Review

(January)
10(1) :159-187.

Freedman, Ronald (ed.). 1975. "The Sociology of Human Fertility: An Annotated
Bibliography." A Population Council Book. New York : Irvington Publishers.

Gibril, M M. 1979. "Evaluating Census Response Errors: A Case Study for the
Gambia," Development Centre of the Organization for Economic Corporation and
Development, Paris, 1979.

Glass, David. 1968. "Fertility Trends in Europe since the Second World War."
Population Studies (March) 22:103-46.

Goidstein, S. 1972. "The Influence of Labor Force Participation and Education on
Fertility in Thailand." Population Studies 26:419-436.

Goliber. T. 1985. "Sub-Saharan Africa: Population Pressures on Development."
' Population Reference Bulletin 40(l):'Washington D.C.: Population Reference Bureau,
Inc. ‘ ' '

. 1989. "Africa's Expanding Population: Old Problems, New Policies."
Population Reference Bulletin 44 (3): Washington DC: Population Reference Bureau.
Inc.

Gray, Ronald.‘ 1983. The Impact of Health and Nutrition on Natural fertility. In Rodolf
A Bulatao and Ronald D. Lee eds. with Paula E. Hollerbach and John Bongaarts,
"Determinants of Fertility in Development Countries," vol 1, " Supply and Demand for
Children." Report of the National Research Council Committee on Population and
Demography, Panel on Fertility Determinants. New York Academic Press, 139-162.

Higgens, Edward. 1964. "Differential Fertility Outlook and Patterns among Major
Religious Groups in Johannesburg." Social Compass 11(1):23-62.

Hoffman, W. L., and J. D. Manis. 1979. "A New Approach to the Study of Fertility."
Journal of Marriage and the Family 41 (3): 353-96.

. ' 1986. "Child F osterage and High Fertility Interrelationship in Africa." In
Abegac (ed). The Economic Demography of Mass Poverty: Studies in Third World
Societies No. 2a. William, Virginia: William and Mary College.

 

200

Jacoby, Hanan G. 1995. "The Economics of Polygny in Sub-Saharan Africa: Female
Productivity and the Demand for Wives in Cote d'Ivoire." Journal of Political Economy
103(3):938-974.

Jaeger, M. and J .L Pennock. "An Analysis of Consistency of Response in Household
Survey," Journal of the American Statistical Association, vol. 56.

Kasarda, J. D., J. Billy and K. West. 1986. "Status Enhancement and Fertility." Orlando,
Florida: Academic Press.

Keyfitz, Nathan, and Wilhelm Fiieger. 1990. "World Population Growth and Aging."
Chicago, Illinios: Uniiversity of Chicago Press.

Kritz, M.M. 1989. "Women's Status, Education and Family Formation in Sub-Saharan
Africa", International Family Planning Perspectives 15(3);100-105.

Kohl, Robert Allen. 1984. "An Investigation Into the Economic Determinants of
Fertility." American University Studies. Peter Lang Publishing Inc., New York.

Kuzrlet, Simon. 1967. "Population and Economic Growth." Proceedings of the American
Philosophical Society 111, no. 3: 170-193.

Lam, David, Guilbeme Sedalcek and Suzanne Duryea. 1992. "Increases in Women's
Education and Fertility Decline in Brazil." Population Studies Center Research Report
No. 92-255. University of Michigan, Ann Arbor.

Lesthaeghe, R. (ed.) 1989. "Reproduction and Social Organization in Sub-Saharan
Africa." Berkeley: University of California Press.

Lucas, D. 1976. "The participation of women in the Nigerian labor force since the 19505
with particular reference to Lagos." Ph.D. Thesis, London School of Economics and
Political Science. . " ‘

MaciOnis, John J. 1997. Sociology. Prentice Hall Upper Sadd River, New Jersey.

Maddala, GS. 1983. Limited-Dependent and Qualitative Variables in Econometrics.
Econometric Society Monographs No. 3. Cambridge University Press.

Marks, ES. and WP. Mauldin. 1950. "Census Response Errors," Journal of the
American Statistical Association vol. 45.

Matras, Judah. 1973. "On Changing Matchmaking, Maniage, and Fertility in Israel:

Some Findings, Problems, and Hypotheses." American Journal of Sociology
(September) 79:364-87.

201

Mazur, Peter D. 1967. "Fertility among Ethnic Groups in the USSR." Demography
4(1): 172-95.

McIntosh C. Alison, and Jason L. F inkle. 1995. "The Cairo Conference on Population
and Development: A New Paradigm?" Population and Development Review 21. no. 2.

Menken, J. J. Trussell and S. Watkins. 1981. "The nutrition fertility link: An evaluation
of the evidence." Journal of Interdisciplinary History, 11:425-441.

Montgomery, Mark and Aka Kouame and Raylynn Oliver. 1995 "The Tradeoff between
Number of Children and Child Schooling : Evidence from Cote d'Ivoire and Ghana,"
LSMS Working Paper No.112.. World Bank, Washington, DC.

Muhsarn, H.V. 1956. "Fertility of Polygamous Marriages," Population Studies, 10 (1) :3-
16.

Murthy, MN. 1963. "Assessment and Control of Non-Sampling Errors in Census and
Survey," Sankhya (Series B), vol. 25 parts 3 and 4.

Myers, Robert A. 1989. Nigeria. World Bibliographical Series, vol 100 . CLIO Press

Myers, Robert. 1940. "Errors and Bias in the Reporting of Ages in Census Data," in
Transaction of the Actuarial Society of America October-November.

Nag, Moni. 1983. The Impact of Socio-Cultural Factors on Breastfeeding and Sexual
Behavior. In Rodolf A Bulatao and Renald D. Lee eds. with Paula E. Hollerbach and
John Bongaarts, "Determinants of Fertility in Development Countries," vol 1, " Supply
and Demand for Children." Report of the National Research Council Committee on
Population and Demography, Panel on Fertility Determinants. New York Academic
Press, 163-198.

Nag, Moni H., E.G Stockweli- and L.M. Snavley. 1973. "Digit Preference and
Avoidance in the Age Statistics of some recent Afiican Censuses: Some Patterns and
Correlates." International Statistical Review, 41(2): 165-174.

Narnfua, Pelad Paschal, 1981. "Polygyny in Tanzania: Its Determinants and Effect on
Fertility." Ph.D Dissertation. The John Hopkins University, Baltimore, Maryland.

National Research Council. 1993. Demographic Change in Sub-saharan Africa. National
Academy Press.

. 1993. Factors Affecting Contraceptive Use In Sub-Saharan Africa.
National Academy Press.

 

202

NewAfrica Year Book 1997-98. Alan Rake (ed.). IC Publications Ltd. 7 Coldbath
Square, London, United Kingdom

Nixon, J .W. 1963. "Some Demographic Characteristics of Protestants and Catholics in
Switzerland," International Population Conference, 1961. No. 2. New York.

Oaxaca, Ronald. 1973. "Male-Female wage differentials in urban labor markets."
International Economic Review 9, 693-709.

Obadike, PD. 1968. "A demographic note on marriage, family and family growth in
Lagos, Nigeria," In J .C. Caldwell and C.N Okonjo (eds.) The Population of Tropical
Africa. Longmans.

Okojie, Christiana EB. 1991. "Fertility Response to Child Survival in Nigeria: An
Analysis of Microdata from Bendel State". in T. Paul Schultz (ed), Research in
Population Economics 7(4). Greenwich, CT: JAI Press.

. 1990. "Women's Status and Fertility in Bendel State of Nigeria."
Economic Growth Center Discussion Paper No. 597. New Haven, Connecticut: Yale
University. '

 

' Orji, Edward. 1983. Marriages Are Made. Printed in Nigeria by Jos University Press
Limited.

Orubuloye, 1.0 1981. Abstinence as a Method of Birth Control : Fertility and Child-
Spacing Practice Among Rural Yoruba Women of Nigeria. Changing African Family

Project Series, Monograph No. 8. The Department of Demography. The Australian
National University. Canberra.

Owusu, J.Y. 1985. "Fertility Preference and Utilization of Family Planning Services." In
Singh, S., Owusu J. Y. and Shah, H. 1985. (eds). Demographic Patterns in Ghana.
Evidence from the Ghana Fertility Survey 1979/80. Voorburg, Netherland International
Statistical Institute.

Oyewole, A. 1987. Historical Dictionary of Nigeria. Africa Historical Dictionaries No.
40. The Scarecrow Press, Inc. Metuchen, N.J., & London.

Pebbley, A.R., and W. Mbugua. I989. "Polygyny and Fertility in Sub-Saharan Africa."
In Lesthaeghe (ed) (1989).

Pitt, Mark. 1994. "Women's Education, Selective Fertility and Child Mortality in Sib-
Saharan Africa". LSMS Working Paper .119. World Bank, Washington.

Populations Reference Bureau. 1997. 1997 World Population Data Sheet.

203

. 1996. 1996 World Population Data Sheet.

 

Potter, J. E. 1971. "Problems in Using Birth History Analysis in Estimating Trends in
Fertility. Population Studies 31 (2): 335-364.

Preston, Samuel. 1978. "The Effect of Infant and Child Mortality on Fertility," New
York Academic Press.

Reinis, Kia L., Shea O. Rutstein and 0.0. Ajaya. 1991. "Fertility Decline in Nigeria : Is
it real or is it Memory Lapse?". A paper presented at the Population Association of
America's Annual National Conference, 1991.

Rizk, Hanna. 1973. "National Fertility Sample for Jordan, 1972: The Study and Some
F indings," Population Bulletin (UNESOB) (July) 5:14-31

. 1963 "Social and Psychological Factors Affecting Fertility in the United
Arab Republic," Marriage and Family Living 25 (1) : 69-73.

 

Rodirguez, German and John Cleland. 1981. "Socioeconomic Determinants of Marital
Fertility in Twenty Countries: A Multivariate Analysis." In World Fertility Survey
Conference 1980: Record of the Proceedings, London, 7-11 July 1980. Voorburg, The
Netherlands: International Statistical Institute, vol 2: 337-414.

Rosenzweig, Mark R. and T. Paul Schultz. 1985. "The Demand for and Supply of Births:
Fertility and its Life Cycle Consequences". American Economic Review 75 (5).

Rutstein, Shea Oscar and George T. Bicego. 1990. "Assessment of the Quality of data
used to Ascertain Eligibility and Age in the Demographic and Health Surveys." In An
Assessment of the DHS-I Data Quality, 5-40. IRD/Macro Systems. DHS
Methodological. Reports No.1 Columbia.

Ryder, Norman B. and Charles F. Westoff. 1971. Reproduction in the US: 1965.
Princeton: Princeton University Press.

Sabagh, Georges and Christopher Scott. 1965. "An Evaluation of the use of
' Retrospective Questionnaires for obtaining vital data: The Experince of Mexican

Multipurpose Sample Survey of 1961-63," United nations World Population Conference
Belgrade (WPC/WP/56).

Schultz, T. Paul. 1997. "Demand for Children in Low Income Countries." Chapter 8 in
Handbook of Population and Family Economics. vol 1A. Mark R. Rosenzweeig and
Odeb Stark eds. Printed in the Netherlands by Elsever Science B.V.

. 1981. Economics of Population. Reading, Mass: Addison-Wesley
Publishing Company.

 

204

. 1974. Economics of the Family. Chicago: University of Chicago Press.

 

Scott, Chris and V.C Chidambaram. 1985 World Fertility Survey: Origins and
Achievements. In John Clelenad and Hobcrafi. Reproductive Change in developing
Countries: Evidence from the World Fertility. New York: Oxford University Press. pp: 7-
26.

Scott, C. and G. Sabah. 1970. "The Historical Calendar as a Method of Estimating
Age,"Popu1ation Studies vol. 24 (I).

Scribner, Susan. 1994. Policies Affecting Fertility and Contraceptive use: An assessment
of 12 Sub-Saharan African Countries. World Bank Discussion Paper No. 259. Policy
Research Department, Washington, DC. Seltzer, W. 1973. "Demographic Data
Collection -- A Summary of Experiences,‘ The Population Council.

Sembajwe, I.S.L 1981. "Fertility and Infant Mortality Amongst the Yoruba in Western
Nigeria." Department of Demography, Australian national University. Canberra.

Shapiro, David. 1996. " Fertility Decline in Kinshasa." Population Studies 50: 89-103.

Shapiro, D and B0. Tambase. 1994. "The Impact of Women's Employment and
Education on Contraceptive use and Abortion in Kinshasa, Zaire." Studies in Family
Planning, 25 :96-110.

Shryoch, Henry 8., Jacob S. Siegel and Associates. 1971. The Methods and Materials of
Demography, vol 1. Washington DC. U. S. Department of Commerce, Burea of the
Census.

Sinha, J .N. 1957. "Differential Fertility and F amiiy Limitation in the Urban Community
of Uttar Pradesh." Population Studies (November) 11 :157-69.

Snyder, Donald W. 1974. "Economic Determinants of Family Size in West Africa".
Demography 11(4): 613-627

Srinivasan, TN. 1994. "Data Base for Development Analysis: An Overview," Journal of
Development Economics. June 44 (1) pp: 3-27.

Stockwell, E.G. 1996. "Patterns of Digit Preference and Avoidance in the Age Statistics

of some Recent Nation Censuses: A Test of the Turner Hypothesis." Eugene Quarterly,
13(3): 202-208.

Standing, G. 1983. " Women's Work Activity and Fertility." In R. A. Bulatao and R. D.

Lee (eds). 1983. Determinants of Fertility in Developing Countries. New York Academic
Press. pp. 416-438

205

Strauss, John and Duncan Thomas. 1996. "Measurement and Mismeasurement of Social
Indicators," American Economic Review. May, vol. 86 (2).

Stycos, J. and R. Weller. 1967. "Female Working Roles and Fertility." Demography 4
(1) 210-217.

Thomas, Duncan, 1995. "Fertility, Education and Resources in South Africa." Labor and
Population Program Working Paper Series 96-15.

Thomas, Duncan and John Maluccio. 1996. "Contraceptive Choice, Fertility, and Public
Policy in Zimbabwe." World Bank Economic Review (January) 10 (l) : 189-222.

United Nations Demographic Yearbook, 1962.

United Nations. 1987. "Education and Fertility." in Fertility Behavior in the Context
of Development. United Nations 1987.

van de Walle, Etienne and Andrew Foster. 1990. "Fertility Decline in Africa:
Assessments and Prospects." World Bank Technical Paper No. 125. World Bank,
Washington, DC.

. 1968a: "Characteristics of African Demographic Data," in Brass et al. The
Demographic of Tropical Africa. Princeton.

 

. 1968b. "Note on the effects of Age Misreporting." In Brass et al. The
Demography of Tropical Africa: Appendix B to chapter 3. Princeton, NJ. Princeton
University Press.

Ware, H. 1979. "Language Problems in Demographic F ieid Work in Africa: The case of
the Cameroon Fertility Survey," WFS Scientific Report No. 2 October, London, World
Fertility Survey.

. 1975. "The Limits of Acceptable Family Size in Western Nigeria."
Journal of Biosocial Science, 7:273-296.

 

Westoff, Charles F. 1959. "Religion and Fertility in Metropolitan America," in Thirty
Years of Reserach in Human Fertility: Retrospect and Prospect. Annual Conference of
Milbank Memorial Fund, October 22-3, 1958. New York: Milbank Memorial Fund.

Willis, Robert J. 1973. "A New Approach to the Economic Theory of Fertility
Behavior," Joumai of Political Economy (Mar/April) 81 (2): 814-864.

World Bank. 1997. World Development Report: The State In a Changing World. New
York: Oxford University Press.

206

World Bank, 1984. World Development Report 1984. New York: Oxford University
Press.

. 1992. World Development Report 1992: Development and the
Environment. New York : Oxford University Press.
. 1986. Population Growth and Policies in Sub-Saharan Africa. Washington,

 

 

D.C.

World Fertility Survey, 1984. World Fertility Survey: Major Findings and Policy
Implications: London: 181.

Yankey, David. 1961. Fertility Differences in a Modernizing Country: A Survey of
Lebanese Couples. Princeton: Princeton University Press.

207

"I111111111111111111113