African Population Studies/Etude de la Population
Africaine, Vol. 19, No.
2, August 2004, pp. 139-157
Effects of Womens Schooling on Contraceptive
Use and Fertility in Tanzania
Ayoub S. Ayoub
Research Statistician, University of Nevada School
of Medicine, University of Nevada Las Vegas
Code Number: ep04016
ABSTRACT
This study explores the economic relationships between
womens schooling, fertility rates, and contraceptive use in Tanzania where
population growth and fertility rates are among the highest in the world
and aggravate the already ailing economy. Two models are used: fertility
and contraceptive use. This study covers women ages 15 to 49. Drawing on
1996 data from the Demographic and Health Surveys (DHS), the study finds
that womens schooling and other socioeconomic variables are important
in explaining reproductive behavior. The fertility model indicates that
higher education levels are consistently associated with lower fertility
rates. Likewise, the contraceptive use model indicates that more education
is positively associated with contraceptive use. Both models show that
the relations become stronger with higher levels of schooling. The findings
indicate that raising womens education levels improves their economic
opportunities, increasing the value of their time and, in turn reducing
their desire for large families.
INTRODUCTION
On October 12th, 1999, the United
Nations announced that global population had reached the 6 billion mark,
just 12 years after passing 5 billion (World Population Data Sheet, 2002).
Based on the United Nations Population Divisions most recent projections,
the global population could reach the 7 billion mark as early as 2011 or
as late as 2015. Most of the increases in population growth can be attributed
to developing countries especially in Sub-Saharan Africa (SSA) countries
where fertility rates are very high. Even though population trends remain
difficult to predict, it is beyond doubt that understanding global population
projections requires an analysis of fertility rates.
High fertility rates could be one of the major
deterrents to sustained economic growth in SSA countries. The ill-effects
of population growth can be examined at macro and micro levels. At a macro
level, high population growth combined with stagnant income can result in
growing income inequalities, lack of economic opportunities and high level
of unemployment. In SSA countries where productivity level is low, food production
cannot keep up with population growth, which leads to food insecurity. SSA
countries are predominantly agricultural based which puts pressure on land
used. Densely populated area results in limited arable land for production
and consumption.
Another problem created by high population growth
is congestion and rapid depletion of resources, especially in developing
countries where property rights governing access to the resources are not
well-defined. This leads to overexploitation of resources, pollution, and
degradation of the environment. Moreover, pressure on limited land availability
in the rural areas due to high population growth has contributed to a massive
migration of peasants to urban centers. Indeed, migration to the city has
led to the mushrooming of slums in the cities, which has exacerbated the
problems of unemployment, lack of proper hygiene, and education opportunities.
At the micro level, high population growth leads
to a more serious issue of poverty. Poorer families, especially women and
marginalized groups, bear the burden of a large number of children with fewer
resources per child, further adding to the spiral of poverty and deterioration
in the status of women. Low levels of income among the poorer families with
many children leads to inadequate food availability, which perpetuates malnutrition,
which in turn accelerates high levels of infant and maternal morbidity and
mortality. Studies by Ernst and Angst (1983), Rodgers (1984), and King (1985)
have widely reviewed the relationship between family size, mean education
and the health of children. Among poorer families, beyond a certain family
size, additional children are usually associated with lower average educational
attainment and reduced levels of child health as measured by nutritional
status, morbidity and mortality.
Problems posed by high fertility rates and population
growth in developing countries have sparked studies of the factors determining
fertility rates. Holsinger and Kasarda (1976), Cochrane (1979) and (1983),
Graff (1979), United Nations (1987), Cleland and Rodriguez (1988), Jejeebhoy
(1992) and (1995) have examined the relationship between female education
and fertility. Generally, these studies found that fertility fell uniformly
with increased levels of womens education. Moreover, in their research in
SSA and Latin America, Jejeebhoy (1995) and Martin (1995) showed that the
inverse relationship between education and fertility can be enhanced only
after relatively high levels of education have been attained.
This study explores the relationships between
womens schooling, fertility rates, and contraceptive use in Tanzania. The
choice of Tanzania has been facilitated by the fact that previous research
has tended to aggregate observations from many countries, and not much has
been written specifically about Tanzania. Also, Tanzania is the largest
of the East African nations with as much cultural and economic diversity
as can be found in almost the entire region. Therefore, the results could
be similar and indicative of the whole region. Results may also be different
and specific to Tanzania. It is important for policy makers to know how Tanzania
is similar to or different from other parts of Africa.
In exploring the effects of womens schooling on
fertility and contraceptive use in Tanzania, two components will be examined.
One component of the study measures the probability of controlling fertility
rate due to womens schooling. I propose that womens schooling levels will
eventually lead to lower fertility rates in Tanzania. The second component
of the study will measure the probability that womens schooling will lead
to contraceptive use.
Like most of previous studies in the analysis
of fertility, this study uses the field data from Demographic and Health
Survey (DHS) conducted in 1996. The Demographic and Health Surveys program
is funded by USAID and implemented by Macro International Inc. MEASURE DHS+ assists
developing countries worldwide in the collection and use of data to monitor
and evaluate population, health, and nutrition programs. The survey data
provide information on family planning, maternal and child health, child
survival, HIV/AIDS/STIs (sexually transmitted infections), and reproductive
health.
Two models are used in the analysis. First, a logit
model is used to estimate the relationship between womens schooling levels
and contraceptive use. The second model is the negative binomial regression
that estimates the probability that increasing womens schooling levels lowers
fertility rates in Tanzania. This study investigates two hypotheses concerning
womens reproductive behavior: (1) that more educated women exhibit lower
fertility, and (2) more educated women are more likely to use contraceptives.
LITERATURE REVIEW
There is a large theoretical literature on the
relationships between female education, fertility, and contraceptive use.
Generally, the results are consistent with predictions of utility theory,
showing that women with more schooling behaved rationally when considering
their family sizes by having fewer children. However, there is little empirical
studies tying together womens schooling, fertility, and contraceptive use.
While advancing the understanding of the determinants of fertility and contraceptive
use, previous studies have focused on only a few variables. For example,
they have neglected to examine the role of other important factors such as
cultural traits in fertility and contraceptive use decisions.
The Relationship between Education and Fertility
The association between education and fertility has a
long history in the fields of economics and demography. Numerous studies
relating national or regional levels of education and fertility showed a
significant inverse relationship between the two. Cochrane (1979) argues
that earlier economists such as Malthus and his successors have proposed
theories about why more education is inversely related with fertility. However,
the relationship between education and fertility is much more complex than
suggested. Though the underlying pattern most commonly known shows a negative
relationship, there are instances where positive relationships at very low
and very high levels of schooling have been found. Bledsoe, Johnson-Kuhn
and Haaga (1999) suggest that understanding the nature and strength of the
relationship between education and fertility remains a central challenge
both for researchers seeking to elucidate demographic and social changes
and for policy makers who must decide on the allocation of scarce public
resources.
The negative effect of education on fertility deserves further
analysis. According to Martin and Juarez (1995, pp. 53), education is a source of
knowledge transmission, vehicle of socioeconomic advancement, and a transformer of
attitudes. In the contemporary world, any development depends on the effective
transmission of new information. As a source of knowledge transmission, Martin
and Juarez discuss that schooling imparts literacy skills, which enable people
to process a wide range of information and arouse cognitive change that shape
individuals interaction with their surrounding environment.
As a vehicle of socioeconomic development, the authors hypothesized
that education not only enhances cognitive abilities, but also it opens up
economic opportunities and social mobility. In the contemporary world, education
credentials open the door for formal employment and for sorting individuals
into the hierarchy of occupations.
Martin and Juarez (1995) explain that as a transformer of
attitudes, schoolings role in attitude formation goes far beyond the enhancement
of conceptual reasoning and may lead to crucial transformations in aspirations
and, eventually, to questioning traditional beliefs and authority of structures.
Education transforms individual attitudes and values from traditional toward
modern and thereby enhancing modernization, which is essential and reliable
to regulate fertility.
Educated women are more likely to exercise the quality-quantity
trade-off of their children. Most of these women are likely to see the benefit
of their schooling; they may develop higher aspirations for their own childrens
schooling. It is obvious that as the number of children increases, familial
resources available to an individual child decrease. Restricting the number
of children is the best solution in order to have better-educated children
and more familial resources per child. It would be advantageous for a woman
to have fewer children that she can afford to pay for the tuition and other
related fees associated with schooling, hence the trade-off between quality
and quantity of children. Ainsworth, Beegle and Nyamete (1996) found that
the trade-off is not a new phenomenon in most of the developed society, but
it is a recent trend that can be seen in some parts of the SSA countries.
Other Determinants
Besides education, a large number of variables can affect
fertility and contraceptive use. For example, Bongaarts, Frank, and Lesthaeghe
(1984) consider two groups of variables: socioeconomic variables and proximate
variables. Socioeconomic variables include education, social, cultural, economic,
and health variables whereas proximate variables include biological and behavioral
variables such as contraception and age of a woman. Davis and Blake (1956)
and Bongaarts and Potter (1983) hypothesize that in order for the socioeconomic
variables to affect fertility, they must operate through proximate determinants.
Cultural traits such as son preference and number
of siblings are important to explain fertility behavior in a traditional
society such as Tanzania, therefore, they deserves to be looked in detail.
Son preference is not an uncommon phenomenon among SSA countries. Khan and
Khanum (2000) found that sons are generally preferred over daughters owing
to a complex interplay of economic and socio-cultural factors. Hank and Hans-Peter
(2000) suggest that son preference is embedded in cultural and religious
traditions and community norms as well as economical factors, shaping individual
attitudes and behavior. In most developing countries where women are economically
and socially dependent on men, male offspring are presumed to have greater
economic net utility than female offspring. The argument is that sons can
help to provide old age support to their parents. This is particularly important
in most developing countries where there is no other form of old-age security.
Hank and Hans-Kohler (2002) suggest that sex preferences for children might
have implications for a couples fertility behavior, where parents who desire
one or more children of a certain sex should tend to have larger families
than would otherwise be the case.
Studies by Duncan, Freedman, Coble and Slesinger
(1965), Axinn, Clarkberg, and Thornton, (1994) have found a direct relationship
between the number of children born to a family and the number of children
within the couples (husband and/or wife) family. In other words, a couple
from larger families is more likely to mimic the sexual behavior of their
parents hence breeding intergenerational inheritance of family size.
METHODOLOGY
The first model analyzes the determinants of contraceptive
use, while the second model deals with fertility. These two models are dependent
upon socioeconomic, demographic and proximate variables. Table 3 in the appendix
gives definition and coding of the variables.
The contraceptive use model, the logit technique
is used because the dependent variable is dichotomous. The logit equation
of the contraceptive use is as given in equation 1:
P(contr) = α0 + α1edprimar
+ α2edsecond + α3knows + α4green
+ α5lnage
+ α6urban
+ α7tv + α8sibl + α9mored
+ ℮1 (eq.
1)
The dependent variable is the probability a woman
uses contraceptive before her first child, while the independent variables
are similar with those of the fertility model, with the exception that in
this model contraceptive use is a dependent variable. Another exception is
the exclusion of higher education variable because almost every woman who
has higher education uses contraceptives.
For the fertility model assumes that women attempt to maximize
their level of utility given all goods and services, including non-market
goods. Accordingly, we specify the following: The negative binomial equation
for fertility is as given in equation 2:
fert = ß0 + ß1edprimar
+ ß2edsecond + ß3edhigher + ß4knows + ß5contr
+ ß6green + ß7lnage
+ ß8urban + ß9tv + ß10mored + ß11Sibl
+ ℮2 (eq. 2)
The dependent variable is the number of children
per a woman. The independent variables include womans schooling levels,
her knowledge of ovulatory cycle, contraception, family planning, age, place
of residence (urban vs rural), income, son preference and number of her siblings.
For the fertility model, this study uses maximum likelihood negative binomial
regression model. Long (1997, pp. 217) posits that the use of linear regression
model for count data can result in inefficient, inconsistent, and biased
estimates. For that reason the negative binomial regression technique is
used in lieu of linear regression technique for the fertility model which
measures count data.
As Ainsworth et al. (1996) hypothesize, all of
the exogenous (except contraceptive use which in the case of the contraceptive
use model is the dependant variable) variables leading women to have fewer
children should result in a positive association in a contraceptive use model.
In other words the coefficients of the independent variables for the two
models will be opposite to each other. Therefore, for simplicity, the following
discussion of the expected signs of the coefficients is centered on the fertility
model. If an independent variable in the fertility model has a positive sign,
the same variable is expected to have a negative sign in the contraceptive
use model and vice versa.
Womens schooling is incorporated in the study
in three distinct ways: primary education level (edprimar), secondary
(edsecond) and higher education (edhigher). No education was
used as a reference variable. It is expected that womens schooling will
have a negative coefficient. Given the opportunity costs of childrearing
(which is time-intensive), the utility of the woman will be maximized by
reducing the number of children to reproduce and spend more time in other
earnings-activities.
The knowledge of ovulatory cycle may also affect
the probability of a woman to have fewer children. This is especially important
in developing countries such as Tanzania where the number of unwanted births
is very high partly because of women not knowing their reproductive cycle.
The variable (know) was measured by asking a woman at what time during
her menstrual cycle she is likely to get pregnant. It is expected that the
variable will have a negative coefficient.
The use of contraceptive is measured by coding
the variable contr 1 if a woman used contraceptive before her first
child; otherwise the variable is coded zero. Contraceptive use is one of
the important determinants of fertility control. The variable is expected
to have a negative coefficient suggesting that efficient contraceptive use
reduces the number of children born per woman.
The variable green is used in this study
to see the influence of family planning knowledge on number of children born
per a woman. The variable is measured by asking women if they ever heard
a local family planning program in Tanzania. It is expected that the variable
will have a negative coefficient to indicate that knowledge of family planning
decreases the number of children a woman may have.
The probability for a woman to have many children
also depends on her age. During the survey women were asked their age. This
study uses lnage which is age in natural log form to reduce the magnitude
of the variable. The survey used in this study covers women aged 15 to 49.
The mean age of the respondent was 28. This means that most of the respondents
were in their childbearing age. Therefore, the coefficient of the age variable
is expected to be positive.
The variable for urbanization (urban)
is 1 if a woman lives in the city. The coefficient on the variable urban is
expected to be negative suggesting that women who live in urban areas will
have fewer children than their counterparts in the rural areas.
The coefficient of variable tv which
is proxy for income is expected to be negative which suggests that women
with relatively high income tend to have fewer children so that they can
have more time to spend on income-generating activities. Another reason is
that women who have more income would rather have fewer children than many
in order to have more expenditure per child which improves the quality of
children as opposed to the quantity of children.
The coefficients of the cultural trait variables;
son preference (mored) and number of siblings (sibl) are expected
to be positive. This suggests that son preference has a higher probability
in increasing the number of children especially when couple gets a child
of a different sex from what they expected. Ceteris paribus the number of
siblings a woman has does have a direct influence on her reproductive behavior,
which means, the more siblings a woman has, the more children she will bear.
Regression Results of the Contraceptive Use
Model
Empirical results regarding the use of contraceptive
reinforce the expectation of the variables in the model. The variables edprimar,
edsecond, knows, green, lnage, and urban are all statistically
significant at the one percent level, while mored is statistically
significant at the five percent level. The variables tv and sibl are
also important but not statistically significant.
The results show that womens schooling significantly impacts
the probability of using contraceptives. The coefficient on the primary education
is positive, indicating that womens primary education increases the chances
of using contraceptive. The marginal effects show that if a woman has primary
education, the likelihood of contraceptive use increases by a factor of 0.043,
holding all other variables constant at their mean. The contraceptive use
model also show that womens secondary education is positive, indicating
that if a woman has secondary education the likelihood of contraceptive use
increases by a factor of 0.073, holding all other variables constant at their
mean. The results of womens schooling on contraceptive use are consistent
with the a priori expectations, contraception is directly associated with
the levels of womens education.
As the results show, womens schooling is positive and statistically
significant at the one percent level on both primary and secondary education
levels. The magnitudes of the coefficients increase with the increase of
womens education level. The results also confirm that womens education
increases receptivity of awareness and contraceptive use to control fertility.
This supports the study done by Bertrand et al. (1993) who found that education
affect the distribution of authority within households, whereby women may
increase their authority with husbands, and affect fertility and use of family
planning.
Table 1: Regression Results: Contraceptive Use Model-Logit
Dependent variable (contr) is 1 contraceptive
used and 0 otherwise
Variable
|
Coefficient
|
Marginal Probability
|
Standard Error
|
Edprimar
|
1.258**
|
0.043
|
0.293
|
Edsecond
|
2.135**
|
0.073
|
0.322
|
Knows
|
0.654**
|
0.022
|
0.143
|
Green
|
0.927**
|
0.032
|
0.138
|
Lnage
|
-1.772**
|
-0.061
|
0.234
|
Urban
|
0.564**
|
0.019
|
0.131
|
Tv
|
0.079
|
0.003
|
0.233
|
Sibl
|
-0.015
|
-0.001
|
0.023
|
Mored
|
-0.393*
|
-0.013
|
0.162
|
Intercept
|
0.523
|
|
0.794
|
Notes: **significant at 1 percent level. *significant
at 5 percent level.
Correct knowledge of a womans ovulatory cycle
as measured by the variable knows, is positive and statistically
significant at the one percent level. Its marginal effect suggests that if
a woman has knowledge of ovulatory cycle, the likelihood of contraceptive
use increases by a factor of 0.022. This suggests that women who know when
they are likely to conceive are more likely to exercise contraception in
case they do not intend to get pregnant.
The awareness of local family planning program
(green) significantly impacts the probability of contraceptive use.
The coefficient on family planning program variable is positive and statistically
significant at the one percent level. The marginal probability shows that
if a woman is aware of a family planning program, the likelihood of contraceptive
increases by a factor of 0.032, holding all other variables constant at their
mean. This indicates that local dissemination of family planning knowledge
is important to explain contraception. However, as mentioned in the previous
section, the local family planning program was launched just three years
before the survey was conducted. Therefore, the finding should be interpreted
with a caveat because the dependent variable (contraceptive use) was measured
as the use of contraceptive before a first child was born. In the fertility
model, this variable was not significant; its significance in the contraceptive
use model may be due the fact that people adjust more quickly on contraceptive
use than on the fertility issue. Another reason may also be that most of
the family planning programs not only disseminate information on family planning
but also on the sexual transmitted diseases. Therefore, people may opt to
use contraceptive such condom for the sake of having safe intimacy and not
controlling fertility.
The model results indicate that womans age is
a strong indicator of likelihood of using contraceptive. The coefficient
that controls womans age is statistically significant at the one percent
level. Since age (in natural log) is a continuous variable, the resulting
coefficient shows the relation to the average age of respondents in the sample.
The negative effect can be explained by the fact that the mean age of surveyed
women is 28, which means that most of these women were in their childbearing
age. Its marginal effect suggests that ages responsiveness decreases the
likelihood of contraceptive use by a factor of 0.061, holding all other variables
constant at their mean.
The results show that urbanization significantly
impacts the probability that an urban-based woman is more likely to use contraceptive
than the rural based woman. The coefficient on urban is positive,
indicating that an urban resident is more likely to use contraceptive. The
marginal probability shows that being an urban resident increases the likelihood
of, contraceptive use by a factor of 0.019, holding all other variables constant
at their mean. This supports Bertrand et al. (1993) who found modern contraceptive
use is higher in urban the in the rural areas. Hamill, Tsui, and Thapa (1990)
attribute the more use in modern contraceptive among urban couples to the
desire for smaller families. This also suggests that urban women may be more
likely to use contraceptive (especially modern contraceptive methods) than
rural women because of greater access to modern methods and medical care
as well as other social amenities in urban areas.
Household income as measured by the dummy of
asset ownership (tv) is positive which is consistent with a priori
expectations. Its marginal effect indicates that if a woman owns a television
set the likelihood of contraceptive use increases by a factor of 0.003, holding
all other variables constant at their mean. However, the coefficient is not
statistically significant; the reason may be that this is a poor measure
of income. Unfortunately, DHS does not collect data on income or household
resources. This problem of poor measurement of income will persist and as
a result will not be possible to address precisely and with confidence the
effect of income on contraception in otherwise extremely rich DHS survey
data.
The result shows that women with many siblings
are less likely to exercise contraception as shown by the negative sign of
the coefficient sibl. However, the coefficient is not different from
zero, which suggests that there is no evidence of a relationship between
the number of siblings and contraception and the exact reason is far from
clear.
The results also suggest that son preference
is one of the major obstacles in using contraceptives making it difficult
to curb the fertility growth. The coefficient for son preference (mored)
is negative and statistically significant at the five percent level. Its
marginal effect suggests that sons preference decreases the likelihood of
contraceptive use by a factor of 0.013, holding all other variables constant
at their mean. This is consistent with a priori expectation that women who
have a son preference will not use contraceptive as long as they bear only
daughters. They will continue that trend until they are satisfied by having
their desired number of sons. If they do not have a son at all they will
keep on bearing children until they reach menopause.
Regression Results of the Fertility Model
Table 2: Regression Results: Fertility Model-Negative
Binomial
Dependent variable (fert) is the number
of children born per a woman.
Variable
|
Coefficient
|
Marginal Probability
|
Standard Error
|
Edprimar
|
-0.040*
|
0.961
|
0.016
|
Edsecond
|
-0.238**
|
0.788
|
0.040
|
Edhigher
|
-0.560
|
0.571
|
0.398
|
Knows
|
0.076**
|
1.079
|
0.019
|
Contr
|
-0.712**
|
0.491
|
0.069
|
Green
|
0.011
|
1.012
|
0.016
|
Lnage
|
2.500**
|
12.186
|
0.028
|
Urban
|
-0.144**
|
0.866
|
0.019
|
Tv
|
-0.203**
|
0.816
|
0.056
|
Mored
|
0.149**
|
1.160
|
0.015
|
Sibl
|
0.016**
|
1.016
|
0.002
|
Intercept
|
-7.496**
|
0.001
|
0.102
|
R-squared
|
0.583
|
|
|
Adjusted R-squared
|
0.582
|
|
|
Notes: ** significant at 1 percent level. *significant
at 5 percent level.
The results of the negative binomial regression
model as well as marginal probabilities of the coefficients. In general,
the empirical results regarding the fertility model are consistent with existing
findings. The variables edsecond, knows, contr, lnage, urban, tv, mored, and sibl are
statistically significant at the one percent level. The variable edprimar is
statistically significant at the five percent level but the variables edhigher and green are
not statistically significant. However, even though the variable knows is
statistically significant at one percent level, it has an unexpected sign.
Interpretation of the count data model using marginal probabilities makes
more sense (See, Long 1997, pp. 224) because the partial derivative cannot
be interpreted as the change in the expected count for a unit change in
an independent variable.
The results show that womens schooling significantly
reduces the number of children born per woman. All the measures of womens
schooling are strongly significant and the coefficients are negative. The
marginal probabilities of education variables show that if a woman has only
primary education, the expected number of children born decreases by a factor
of 0.961, holding all other variables constant. If a woman has secondary
education, the expected number of children born decreases by a factor of
0.788 and if a woman has higher education, the expected number of children
born decreases by a factor of 0.571, holding all other variables constant.
The results are consistent with previous studies (e.g. Ainsworth et al. (1996),
Martin and Juarez (1995) and Adelman (1963).
Higher education is however, not statistically
significant, but the negative magnitude of the coefficient is higher than
those of womens primary and secondary schoolings. Notice that the negative
effect of womens education on fertility gets larger and larger with the
increase in education levels. The insignificance of higher education coefficient
may be explained by the fact that only 0.05 percent of the surveyed women
have higher education beyond secondary school. More data may be needed to
significantly capture this effect. Unfortunately, such data are not available
from the Tanzania Demographic and Health Survey.
The coefficient of knowledge of ovulatory cycle
(knows) is positive and statistically significant at the one percent
level indicating that if a woman has knowledge of ovulatory cycle, the number
of children born will increase. The marginal effect of the coefficient shows
that if a woman has knowledge of ovulatory cycle the expected number of children
born increases by a factor of 1.079 holding all other variables constant.
However, this is inconsistent with the model prediction even after performing
two-stage least squares (2SLS) to see if there is a problem with endogeneity
between the variable knows and fertility. The inconsistency might
be due to the fact that only about 16 percent of the women know their reproductive
behavior. Also, it is possible that a majority of those who know their reproductive
cycle use the knowledge to have more children if they prefer a large family
size.
The use of contraceptive also significantly
lowers number of children born per woman. The coefficient on the contraception
variable is negative and statistically significant at the one percent level,
indicating that as contraceptive use increases, the number of children born
per woman decreases. A change [∂P(fert)/∂contr < 0] gives
a decrease in the expected number of children born by a factor of 0.491,
holding all other variables constant. This suggests that women used contraceptive
before their first child was born, they used contraceptive efficiently and
effectively. This finding is consistent with the studies done by Rutenberg,
Ayad, Ochoa and Wilknson (1991) and Bongaarts, Mauldin and Phillips (1990).
These studies found that the increased use of effective family planning methods
is the primary cause of dramatic fertility declines in many developing countries.
Family planning programs are very important in
explaining a countrys fertility rate and are expected to have a negative
effect on fertility. However, the coefficient is positive, but not significantly
different from zero. So the result shows there is no correlation. One possible
explanation for this peculiar finding could be that the local family planning
program was launched officially in mid 1993 and the survey on which this
study is based was conducted just three years later (1996) so the effect
of the plan had not yet materialized. It also makes sense to assume that
when the program was launched, most of the surveyed women had a higher fertility
rate already. This suggests that the benefits of this family planning program
are not supported by the model, but may be supported by more timely data.
Womens age in natural log form is statistically
significant and has a positive sign which is consistent with the earlier
prediction. Given that the mean age of the surveyed women is 28, suggesting
that most of the women are in their childbearing age. The marginal probability
shows that the responsiveness of a womans age increases the expected number
of children born by a factor of 12.186, holding all other variables constant.
The model results indicate that urbanization
significantly impact the probability of reducing the number of children born
per woman. The coefficient of urban is negative and statistically
significant at the one percent level, denoting a negative relationship between
women who live in urban areas and the number of children born per woman.
The marginal probability suggests that if a woman resides in an urban area,
the expected number of children born per woman decreases by a factor of 0.866,
holding all other variables constant. This is supported by Ainsworth et al.
(1996), who determined that urban women have lower fertility than rural women.
The finding is also consistent with Sharlin (1979) who hypothesized that
the population that lives in urban areas is associated with fundamental revolution
in economic basis. The results also support Adelman (1963, pp.322), who found
out that socioeconomic phenomena associated with the urbanization process
tend to reduce birth rates in the long run.
The results show if a woman lives in a household
that has a television set (income proxy) is more likely to have fewer number
of children born. The coefficient for the variable is negative and statistically
significant at the one percent level. The marginal probability shows that
if a woman has television set, the expected number of children born decreases
by a factor of 0.816, holding all other variables constant. This finding
reflects the stronger impact of mothers income on the time cost of children.
The results support Becker (1965, pp 510) who hypothesized that child care
would seem to be a time-intensive activity that is not productive (in terms
of earnings) and uses many hours that could be used at work which is earnings-intensive
activity. Becker predicted that the opportunity cost of childbearing is
higher to higher income families. This association between household income
and number of children born per woman is probably the most important economic
explanation of decreasing fertility rates.
The data also suggest that cultural trait of
son preference represented by mored is also statistically significant
at the one percent level. This indicates that if a woman has son preference,
the expected number of children born increases by a factor of 1.160, holding
all other variables constant. This shows that women who have son preference
will continue to reproduce until they get a desired number of sons. The finding
is consistent with the study of Hank and Hans-Kohler (2002) who suggested
that parents who desire one or more children of a certain sex should tend
to have larger families than would otherwise be the case.
Lastly, the variable sibl also has a
significant impact on the number of children born per a woman. The coefficient
on sibl is statistically significant at the one percent level and
positive indicating that intergenerational inheritance of family size preference
has a direct influence on fertility. The marginal probability indicates that
number of siblings increases the expected number of children born by a factor
of 1.016, holding all other variables constant. This also implies that women
are more likely to mimic the reproductive behavior of their parents. This
is supported by Duncan et al. (1965), Axinn et al. (1994) who determined
the direct relationship between the number of children born to a family and
the number of children within the couples (husband and/or wife) family.
CONCLUSION
This study has examined the effects of womens
schooling and other socioeconomic and demographic factors on fertility and
contraceptive use in Tanzania. The study finds strong support for the negative
correlation between womens schooling and cumulative fertility. The findings
answer very important questions of how womens schooling, which rarely addresses
issues directly relevant to sexual, reproductive and contraceptive behavior,
influences fertility and contraceptive use decisions. Also, the study provides
evidence of an established positive relationship between females schooling
and contraceptive behavior.
The findings also substantiate that other explanatory
variables such as cultural factors are important in dealing with the question
of controlling fertility. As discussed in this study and various other studies,
the indirect effects of females education on fertility and contraceptive
use are much higher than the direct effects. One of the indirect applications
of education is to empower women with decision-making and counteract cultures
and norms that are associated with low prevalence of contraceptive use and
increased fertility rate.
According to Becker (1992, pp.45) womens education
raises their labor participation which in turn raises their earnings, and
hence greater investment in market-oriented skills which increases womens
time value. Failure to recognize the crucial indirect effects of education
on fertility and contraception is to undermine the role of female education.
Still unknown from this study: Does the inter-relatedness
between females schooling, fertility, and contraception depend on the quality
of education? Unfortunately, exogenous measures of the quality of education
are not among those available in the individual DHS data set. The unavailability
of that information, forces this study and previous studies to use either
years of schooling or education levels to capture the impact of education
on fertility and contraception. The quality of education depends on a variety
of factors including a countrys education system. Given this caveat, one
has to be careful in generalizing the results of this paper to other countries.
Additional research is needed to take a closer look at how different education
qualities behave differently in explaining fertility and contraception in
developing countries where there is much diversity in the quality of education.
This study suggests that investment in womens
education should be a practical priority. Investment in primary education
is necessary, but both fertility and contraception models show that the impact
(magnitude) of education increases with education level. Therefore, the
Tanzanian government should also invest heavily on women schooling beyond
primary school. Investment in females education in secondary and higher
education will foster economic growth and also promotes smaller families,
increase modern contraceptive use, and improve child health.
APPENDIX
Table 3: Variables Used for Both Models and their Definitions
Variable
|
Definition
|
fert
|
Total number of children ever born per
woman
|
edprimar
|
1 if a woman has primary education level
0 otherwise
|
edsecond
|
1 if a woman has secondary education level
0 otherwise
|
edhigher
|
1 if a woman has higher education level
0 otherwise
|
knows
|
1 if a woman has a knowledge of ovulatory
cycle
0 otherwise
|
contr
|
1 if a woman use contraceptive before first
child
0 otherwise
|
green
|
1 if a woman ever heard of a local family
planning
0 otherwise
|
lnage
|
a womans age in log form
|
urban
|
1 if a woman lives in urban residency
0
otherwise
|
tv (income proxy)
|
1 if a woman has television set
0 otherwise
|
mored
|
1 if a woman has more daughters over sons
0
otherwise
|
sibl
|
number of siblings a woman has
|
BIBLIOGRAPHY