Proximate determinants of fertility in Ethiopia; an application of revised Bongaarts model

Background World population is growing at about 80 million people each year. Ethiopia is the 12th most populous country in the world. Existing literatures showed that the role of proximate determinants in inhibiting the total fertility has not yet been determined from the DHS data in the country. This study may provide evidence based information regarding the observed changes in total fertility. The objective of this study was assessing proximate determinants of fertility and the role of selected socio-economic variables in influencing fertility in Ethiopia. Methods The EDHS data of 2011 and 2016 were used in our study. A total of 16,515 eligible women included in 2011 and 15,683 in 2016 surveys made up the sample for the study. The roles of each of the four proximate determinants in declining fertility have been determined. The background variables selected for the analysis include: region of residence, educational status, wealth index and place of residence. The Bongaart model is used to explain the observed socio-economic differentials in fertility during the two survey years. Results In 2011, index of marriage inhibited fertility by 37.8%, however in 2016 it inhibited fertility by 34.4%. In 2011, contraceptive use reduced fertility by 28.5% while in 2016 it reduced fertility by 30.7%. The index of postpartum infecundity decreased fertility by 34.7% in 2011 and by 34.5% in 2016. Foetal wastage inhibited fertility by 9.2% in both survey years. The total fertility rate in 2016 was 4.14 whereas the projected total fertility in 2020 will be 3.2 children per woman. Conclusion Among the four proximate determinants of fertility, the contribution of index of marriage was the highest in inhibiting fertility in 2011. On the other hand, the contribution of postpartum infecundability was the highest in inhibiting fertility in 2016. The contribution of the index of contraceptive in inhibiting fertility increased from 28.5% in 2011 to 30.7% in 2016. The index of foetal wastage contributed the least in both 2011 and 2016 survey years. Therefore, strategies have to be designed to promote the contraceptive use and breast feeding practices among the reproductive women.


Plain English summary
World population is growing. The Ethiopia population is also increasing. Fertility rate is an important determinant of population size. In this study the contribution of the proximate determinants of fertility has been determined.
A total of 16,515 (2011EDHS) and 15,683 women (2016 EDHS) were the study participants. The extent to which each of the four proximate determinants contributed in declining the total fertility has been determined. The Bongart revised model was used to explain the observed socio-economic differentials in fertility.
In 2011, 37.8% of births were inhibited due to the effect of delayed marriage & non marriage, whereas its contribution in declining fertility was 34.4% in 2016. The influence of contraception in fertility inhibition was 28.5 and 30.7% below its biological maximum in 2011 and 2016 respectively. Postpartum infecundability decreased the total fertility by 34.7% in 2011 and 34.5% in 2016. The effects of index of foetal wastage were the same i.e. it inhibited only 9.2% of the total births in both survey years. Based on our projection the future total fertility of Ethiopia in 2020 will be 3.2.
In summary, index of marriage and postpartum infecundability were the leading fertility inhibiting factors in 2011 and 2016 respectively. The contribution of the index of contraceptive in inhibiting fertility increased from 28.5% in 2011 to 30.7% in 2016 whereas foetal wastage contributed the least in both survey years. Therefore, contraceptive use and breast feeding practices have to be promoted among the study subjects.

Background
In demography, fertility indicates the product or output of reproduction, rather than the ability to have children.
The physiological ability to have children that is manifested roughly in the period between menarche and menopause in women is termed fecundity. Demographers define a third, further aspect of reproduction, fecundability, which is the probability of becoming pregnant, or the likelihood of exposure to the possibility, that depends on the pattern of sexual and pregnancy preventive behaviours [1].
Demographically observed fertility or infertility is the result of a well-defined number of both biological and behavioural factors, which serve to mediate the influence of culture, society, economic conditions, living standards, and other similar background determinants on individual reproductive behaviour. Together, these biological and behavioural factors are called the proximate determinants of fertility and they are the factors through which, and only through which, the social and economic environment can influence individual procreation. Together, these factors constrain fertility, and explain why women do not have the maximum possible number of children, which would be about 35 if they reproduced continuously from, say, the age of 18, to about the age of 45, having a birth every 9 months [1].
World population is growing at about 80 million people each year. If current trends keep on, there will be about 9.2 billion people on this planet by the mid of 21st century. Virtually all of the major problems that threaten the world today are connected with rapid population growth. It fuels health risk, hunger or poverty, water scarcity, environmental depletion and political instability [2]. Ethiopia is the 12th most populous country in the world [3] with a recently estimated total fertility rate [TFR] of 4.6 children per woman [4].Though, Ethiopia has achieved remarkable declines in fertility rate, the changes are not shared equally among the different administrative regions of the country [5]. Some regions have shown a fall in fertility, whereas some are moving to the opposite direction and the other regions still have shown no change in total fertility rate [6].
The TFR of Ethiopia declined from 5.5 children per woman in 2000 to 4.6 children per woman in 2016 [4,7]. This puts the country at the rank of 14th highest total fertility rate in the world [4]. The TFR among women declined from 6.0 children in 2000 to 5.2 children in 2016 and 3.0 children in 2000 to 2.3 children in 2016 in rural and urban areas respectively [4,7].
Studies done in different countries identified various diffentials of fertility. Study conducted in Bangladesh indicated contraceptive practice as a key factor in fertility change [8]. In Zambia, marriage and postpartum infecundity accounted for the largest inhibiting effect on natural fertility from its biological maximum of 19.10 [9]. In Philippines, fertility is differentiated by ethnic group, place of residence, educational attainment, women's work and age at marriage. Contraceptive use and wealth index also played an important role in explaining the fertility differentials [10]. The findings from Ghana indicated a woman's contraceptive behavior; marriage status and postpartum infecundability as important predictors of fertility outcomes. Higher education levels and urban residence were consistently associated with lower fertility rates [11]. Most of the decline in fertility in Sub-Saharan Africa is attributable to increases in the proportions of women unmarried and to a lesser extent increases in contraceptive use [12]. Study done in Sub-Saharan Africa, identified proximate determinants of fertility such as non-marriage, post-partum amenorrhea and contraceptive use as an important factors in inhibiting fertility. The Study also identified the background variables such as desired number of children, national family planning effort, under-five mortality, level of education, female participation in the workforce, and place of residence as influencing factors of fertility [13]. In Asian countries, marriage pattern and induced abortion played an important role in reducing fertility among women [14].
John Bongaarts early framed a model explaining fertility as a function of proximate determinants of fertility through which the background variables affect fertility [15]. Revised version of John Bongaarts' model proposed by John Stover [16] is currently in use to indicate the roles of the proximate determinants in predicting the TFR [17]. Our research used a revised Bongaarts model to explain the contribution of each proximate determinant in the total change of fertility. These proximate determinants include exposure to childbearing (marriage, consensual union and sex), contraceptive use, total foetal wastage and postpartum infecundity [16]. Getting abortion data is difficult in Ethiopia. Therefore, we used the revised Bongaarts model to include foetal wastage instead of abortion. The total fertility as a key demographic indicator has been declining. But the contribution of various reproductive behaviours and background variables to this change need to be timely assessed. This helps to determine how much they have inhibited fertility rate over time. The role of background factors in influencing the proximate determinants of fertility and the extent to which proximate determinants influence the fertility decline has to be investigated from the most recently available database. The four background variables: region of residence, educational status, wealth index and place of residence were preferred to use based on their effect on influencing fertility [8][9][10][11][12][13][14][15]. Existing literatures showed that the role of proximate determinants in inhibiting the total fertility has not yet been determined from the DHS data in the country. This study may provide evidence based information regarding the observed changes in total fertility between the two recent DHS years, EDHS 2011 and 2016, and the independent contribution of the proximate determinants to these changes. It also furnishes useful baseline information for further studies which will prospectively make use of the EDHS database. Therefore, the objective of this study was assessing proximate determinants of fertility and the role of selected socio-economic variables in influencing fertility in Ethiopia.

Data sources
The source of data for our study was the EDHS conducted in 2011 and 2016. A total of 16,515 eligible women enrolled in 2011 and 15,683 in 2016 surveys made up the sample for this study. The roles of each of the four proximate determinants in fertility decline was determined. One important fertility determinant variable was not captured by these datasets namely abortion due to lack of appropriate data. Instead of abortion data, foetal wastage data was included in the analysis. The selected background variables include: region of residence, educational status, wealth index and place of residence.

Data analysis methods
John Bongaarts framework for analysing proximate determinants of fertility model was used to analyse the data. The four proximate determinants of fertility, total fertility and background variables were calculated by using Microsoft excel software. In the following section we presented detailed analysis of fertility differential model that quantifies the negative and positive effects of each of the socio-economic and cultural factors on fertility through various intermediate fertility variables. For illustration, the model was used to explain the observed socio-economic differentials in fertility during the 2011 and 2016 EDHS.

Bongaarts fertility differential model
Bongaarts model of estimating the effect of four proximate determinants assumes that the natural reproductive capacity that is total fecundity rates (TF) is nearly the same for all women, but their actual reproductive performance is modified by four major proximate determinants. In his model [7], Bongaarts expressed TFR as the product of four indices measuring the fertility inhibiting effect of these four indices and is the average number of live births expected among women who during their entire reproductive period, remain married, do not use contraception, do not have any induced abortion and do not breastfeed their children [17]. According to Bongaarts model, the TFR can be written as Where, Cm is the index of proportion married, Cc is the index of non-contraception, Ca is the index of induced abortion, Ci is the index of lactational infecundability and TF is the total fecundity.

Revised Bongaarts model
This revised model replaced only the index of all foetal wastage (sum of miscarriage, abortion and still birth) instead of the index of abortion. Other three indices have remained the same as proposed by Bongaarts original model [18]. The revised Bongaarts model [8] is represented as TFR is the effect of all four proximate determinants of fertility. It is calculated by multiplying the four indices. Cm, Cc, Cfw, Ci are the indices of marriage, contraception, foetal wastage, and postpartum infecundability respectively. The indices can only take values between 0 and 1. When there is no fertility-inhibiting effect of a given intermediate fertility variable, the corresponding index equals 1, and if its fertility inhibition is complete, the index equals 0. These indices can be estimated from measures of the proximate variables and their estimates are shown below.

Estimation of the index of marriage (cm)
The index of marriage measures the inhibiting effect of marriage on fertility in the population. It has to be noted that the higher the level of marriage in the population the lower the fertility inhibiting effect and the vice versa. The index equals one when all women are cohabitating, and zero when no women are in a union. The index of marriage is estimated using the formula; Where m (a) = age specific proportions currently married among females, g (a) = age specific marital fertility rates. m (a) is obtained by dividing the number of married women of a particular age group by the number of women in the same age group whereas; g (a) is calculated by dividing the births of a particular age group by the number of women in the same age group.

Estimation of the index of contraception (cc)
The index of contraception in the model measures the inhibiting effect of contraception on fertility in the population. The higher the level of contraception in the population, the higher the inhibiting effect due to contraception and the lower the level of contraception the lower its inhibiting effect on fertility. The Cc equals one when no contraception is used, and zero when all fecund women use 100% effective contraception. The index of contraception is estimated using the formula Where; u = Proportion using contraception among married women of reproductive age (15-49 years). e = Average use effectiveness of contraception. e (m) = method specific use effectiveness, U (m) = proportion of women using specific method. The coefficient 1.08 represents an adjustment for the fact that women do not use contraception if they know that they are sterile. The indices of use effectiveness proposed for particular contraceptives are; pill = 0.90, IUD = 0.95, sterilization = 1.00, injection = 0.99 and others = 0.70 [17] Estimation of the index of foetal wastage (Cfw) The index of fetal wastage (Cfw) is defined as the ratio of the observed TFR to the estimated TFR with all fetal wastage which is (TFR + AFW), that is Cfw = TFR/TFR + AFW. Where AFW equals the average number of births averted per woman by the end of the reproductive years for all foetal wastage and which is estimated by Where, TFW is the total foetal wastage rate and AFW is the average number of all foetal wastages per woman at the end of the reproductive years.

Estimation of the index of postpartum infecundability
The index of postpartum infecundability measures the inhibiting effect of breastfeeding or abstinence on fertility in the population. The index of postpartum infecundability in the model is estimated using the effect of breastfeeding (lactation amenorrhea) or postpartum abstinence. The ratio of natural fertility in the presence and absence of postpartum infecundability therefore equals the ratio of the average birth interval without and with postpartum infecundability. If no breastfeeding and postpartum abstinence are practiced, the birth interval averages about 20 months, which is the sum of; i). 1.5 months of minimum postpartum anovulation. ii). 7.5 months of waiting time to conception. iii). 2 months of time added by spontaneous intrauterine mortality. iv). 9 months for a full term pregnancy.
Bongaarts Stated that, in the presence of breastfeeding and postpartum abstinence, the average birth interval equals approximately 18.5 months (7.5 + 2 + 9) plus the duration of postpartum infecundability [17]. The index of postpartum infecundability (Ci), therefore, is estimated as; Where i = average duration of postpartum amenorrhea. The index is calculated as the average birth interval in the absence of breastfeeding divided by the average length in the presence of breastfeeding. The index equals one in the absence of lactational amenorrhea or postpartum abstinence and declines in size as the period of postpartum infecundability rises.

Results
Estimated indices for the four major proximate determinants of fertility and their impacts According to our analysis the value for Cm in 2011 was 0.622, where both delayed marriage and non-marriage inhibited fertility by 37.8%. In 2016 Cm was 0.656, in which case delayed marriage & non marriage inhibited fertility by 34.4%. The value for Cc in 2011 has been estimated to be 0.715, in which case contraceptive use has reduced fertility by 28.5%, while in 2016 the Cc was 0.693, where the use of contraceptives reduced fertility by 30.7% below its biological maximum of 15.3 births.
The estimated values were presented in Table 1. Our analysis showed that the index for postpartum infecundity (Ci) in 2011 was 0.653, where lactation amenorrhea decreased fertility by 34.7%, and the index for Ci in 2016 was 0.693, where the effect of lactation amenorrhea reduced fertility by 34.5%. The indices for foetal wastages in both 2011and 2016 survey years were the same, i.e. 0.908. In both years foetal wastages inhibited 9.2% of the observed total births. The estimated values were presented in Table 1.

Estimated indices for the four proximate determinants by selected background factors
The contribution of each of the four proximate determinants and their variability by these background variables in declining fertility were presented in Tables 2 and 3.

Index of marriage (Cm)
In our study, in 2011 the highest Cm (0.801) was observed in Afar region whereas in 2016 its highest Cm (0.760) was shown in Somalia region. These showed that delayed marriage and non-marriage reduced fertility by 19

Index of contraceptives (Cc)
According to our analysis, in 2016, the Cc was highest, Cc = 0.984, in Somalia region and lowest, Cc = 0.437, in Addis Ababa region. These signify that contraceptive use inhibited only 1.6% of births in Somalia region whereas 56.3% in Addis Ababa region. The Cc was also the lowest, Cc = 0.377, in Addis Ababa in 2011 which shows that contraceptive use lowered fertility by 63.8%. Similarly, in 2011, the Cc was the lowest, Cc = 0.960, and contributed 4% in fertility reduction in Somalia region. The Cc was analysed in relation to participants' education level and variability has been observed. Accordingly, the lowest Cc, 0.351, was observed in 2011 among women having higher education and the lowest Cc

Index of foetal wastage (Cfw)
The Cfw was calculated using the data from both surveys to detect any differences across the back ground

Total fertility rate
Our analysis was inclusive of the calculation of the total fertility using revised Bongaarts model. As a result, the total fertility rate in 2016 was calculated to be 4.14 children per woman. The findings of our study also showed variability in TFR across the selected background variables. Accordingly, the highest TFR, 8.1 children per woman, and the lowest TFR, 1.5 children per woman, were observed in Somalia and Addis Ababa regions, respectively in 2016. In this year, the highest TFR of 4.9 children per woman was seen in lowest wealth quintile and the lowest TFR of 1.8 children per woman in the highest wealth quintile. In relation to the participants' education, the highest TFR, 2.6 children per woman, and the lowest TFR, 1.8 children per woman, were observed among participants with primary education and higher education respectively in 2016. The TFR was 2.5 in rural and 1.5 in urban areas in this year.

Projection of future fertility
The Bongaarts model of proximate determinants was used to project future fertility of the country. This was analysed by considering the expected contraceptive prevalence rate set by the Ethiopian government to attain a certain fertility level in 2020.
We supposed that TFR1 and TFR2 be the total fertility rates, in year 1 (the present year) and year 2 (year in the future), and the corresponding levels of contraceptive prevalence and use effectiveness be u1 and e1, and u2 and e 2, respectively. We calculated the level of total fertility for the changes of contraceptive uses from year 1 to year 2. However, we assumed that the other proximate determinants except for contraception remain constant, that is Cm1 = Cm2, Ci1 = Ci2 and Cfw1 = Cfw2. It is also assumed that contraceptive use-effectiveness remains constant.
The projected total fertility rate becomes Therefore, the projected TFR in 2020 will be 3.2 children per woman.

Discussion
Our study assessed the proximate determinants of fertility and the role of the selected socio-economic variables in influencing the proximate determinants and fertility in Ethiopia. The revised Bongaarts model was used to determine the effects of proximate determinants on total fertility rate and its variability among the selected background variables. Generally, the results in this study showed that the impact of the four proximate determinants of fertility varies considerably among the selected background characteristics. In 2011, the Cm inhibited fertility by 37.8% whereas in 2016 it inhibited by 34.4%. Consequently, the Cm in 2011 contributed more for the reduction of fertility than Cm in 2016. This variation could be due to the fact that the proportion of married women was larger, 65% in 2016, when compared to 58% in 2011. This could be explained by the fact that the more women of child bearing age engage in marriage, the lower is its fertility inhibiting effect.
Both in 2011 and 2016, Cm played an important role in inhibiting fertility. This is in line with study done in Zambia, which showed marriage as the largest inhibitor on natural fertility [9]. The finding from Ghana also indicated a woman's marriage status as important predictors of fertility outcomes [11]. Studies done in Sub-Saharan Africa identified non-marriage as attributable factor to decline fertility [12,13]. Marriage was seen as the fertility inhibitor in Namibia and in Ethiopia [21,22].The proportion of births inhibited due to the effect of contraception is higher in 2016 (30.7%) when compared to what was seen in 2011 (28.5%). This difference in the inhibition of fertility could be explained by the differences in the contraceptive prevalence rate. The contraceptive prevalence was higher (35%) in 2016 as compared to what was observed in 2011 (29%).e..
In current study, contraception practice played a role in influencing fertility. Study conducted in Bangladesh indicated that contraceptive practice played a key role in fertility change [8]. Contraceptive use also played an important role in explaining the fertility differentials in Philippines [10]. The finding from Ghana, Namibia and Ethiopia indicated that a woman's contraceptive behavior as important predictors of fertility outcomes [11,21,22].
The contribution of contraception in inhibiting fertility in current study is higher than study done in Zambia, which showed that contraception use accounted for only 3% in inhibiting fertility [9]. The proportion of births prevented due to the effect of postpartum infecundability was slightly higher in 2011 (34.7%) as compared to what was in 2016 (34.5%). Postpartum infecundability is the second important proximate determinants of fertility both in 2011 and 2016. This is similar with the findings of Ghana, which indicated that postpartum infecundability was an important predictors of fertility outcomes [11]. Study done in Sub-Saharan Africa, identified post-partum amenorrhea being an important proximate determinant of fertility, which can be substantially extended by prolonged breastfeeding [13]. Study conducted in Namibia and Ethiopia identified postpartum infecundability as an important fertility inhibitor [21,22].
In 2016 survey year the contraceptive use inhibited only 1.6% of births in Somalia region whereas 56.3% in Addis Ababa region. This huge difference observed between the two regions could be explained by variations in contraceptive prevalence rate between the regions, 55.9% in Addis Ababa and 1.5% in Somalia. Moreover, the variation in education among the study participants between the regions could have brought changes in awareness which in turn might have influenced the contraceptive use. Similarly, the highest and the lowest TFR were observed in Somalia and Addis Ababa region, respectively, in 2016. Study conducted in Ghana showed region of residence as predictor of fertility [11].
In 2016, the highest TFR was seen in lowest wealth quintile. Study done in Ghana and Philippines, showed wealth index as predictor of total fertility [10,11].
In relation to the participants' education, the highest TFR and the lowest TFR were observed among participants with primary education and higher education respectively in 2016. This is in line with study conducted in Ghana, which showed higher education levels are consistently associated with lower fertility rates [11]. Study done in Sub-Saharan Africa identified level of education as influencing factor of fertility [13]. In Ethiopia, main differences were observed across educational status [22].
The TFR was lower in urban areas than rural. This is in line with Ghanian study, in which urban residence is associated with lower fertility rates [11]. Study done in Sub-Saharan Africa identified place of residence as influencing factor of fertility [13]. In Ethiopia, main differences were observed across residential environment [22].

Limitation
Reporting and recall bias particularly for age or other retrospective data relying in memory of the past event is the limitation of the study.

Conclusion
Among the four proximate determinants of fertility, index of marriage was the most important inhibiting factor of fertility in 2011. On the other hand, postpartum infecundability was the first important fertility inhibitor in 2016. The contribution of the index of contraceptive in inhibiting fertility increased from 28.5% in 2011 to 30.7% in 2016. The index of foetal wastage contributed the least in both 2011 and 2016 survey years. The study also showed substantial differences in fertility inhibition by background factors in both 2011 and 2016.
The contribution of index of contraceptive in inhibiting fertility rate has not been as expected when compared to the contribution of other proximate determinants. The total fertility rate has shown variations across the regions. Therefore, the government and development partners need to design effective strategies and act in a concerted way to increase the prevalence of contraceptive use across the country, with particular emphasis to infrastructure limited regions.
The contribution of Ci in inhibiting fertility in 2016 was the highest. This calls for further strengthening of breast feeding as a good behavioural practice to promote natural family planning practices among breast feeding women..