Page 61 - IJPS-11-3

International Journal of
Population Studies Age patterns of fertility in Ethiopia

2020). Mathematical models help smooth reported ASFRs, transitions. One of the primary objectives of this study is
project them, and estimate them. The literature includes to analyze the fertility transition in Ethiopia by observing
several parametric and non-parametric mathematical changes over time in ASFRs across various regions.
models that attempt to describe the ASFRs. It is observed Understanding the existing interrelationships and high
that as modernization increases, ASFRs vary across correlations between fertility and mortality measures,
populations based on their stage of demographic transition, socio-economic indicators, and other measures, researchers
leading to the development of new mathematical models have attempted to derive summary indicators such as TFR
(Brijesh et al., 2015; Coale & Trussell, 1974; Gayawan and gross reproduction rate (GRR) using simple regression
et al., 2010). Over the past few decades, fertility patterns models from relevant highly interrelated development and
in developed countries have deviated from the classical demographic indicators (Bogue & Palmore, 1964; Hauer
uni-modal distribution, with a notable bulge in early-age et al., 2013; Jagadeesh Kumar, 1977; Mitra, 1965; Palmore,
fertility observed in ASFR curves. To accommodate this 1978; Premi, 1974; Rele, 1967).
bulge, different models have been developed, assuming
that the distorted fertility distributions may represent Rele (1967) developed a method to estimate GRR and
not a homogeneous population but multiple sub- TFR from the Child–Woman Ratio (CWR) corresponding
populations (Chandola et al., 1999; Peristera & Kostaki, to a given level of life expectancy at birth, based on stable
2007). Recently, researchers like Chao et al. (2023) have population theory principles (Rele, 1967). Bogue &
used the Bayesian hierarchical model (BHM) developed Palmore (1964) estimated various fertility indicators such
by the United Nations Population Division to estimate as crude birth rate, GFR, ASFRs, and TFR from a set of
ASFR trends. The BHM is gaining popularity due to its fertility, mortality, and development indicators. Mitra
ability to incorporate uncertainty and prior knowledge. (1965) attempted to provide a set of regression models
The Kumaraswamy-log-logistic distribution, commonly and a correspondence table to derive ASFRs from GFR
used in survival analysis, is also being used to model information. These researchers generally preferred to use
ASFRs (Gaire et al., 2024). Mishra et al. (2017) used the simple regression models to achieve their goal.
skew-logistic distribution to model India’s ASFR pattern, However, regression models typically provide reliable
which effectively captures the skewed nature of the ASFR estimates only for periods under consideration and
curve. Polynomial models are also used in modeling have limited applicability. Interestingly, Rele’s (1967)
ASFRs (Gaire et al., 2022; Singh et al., 2015). Vanella & methodology, which uses stable population theory to
Hassenstein (2024) proposed a stochastic forecast model estimate GRR/TFR from CWR, still yields plausible results
for regional ASFRs, which can account for demographic in populations currently considered non-stable.
fluctuations and is particularly valuable during times
of economic or political instability when fertility rates Observations from the EDHS survey estimates of TFRs
experience high variability or uncertainty. The integration across various regions in Ethiopia (Figure 1) reveal a
of big data from online sources with machine learning is significant decline in fertility from 2000 to 2016. However,
becoming increasingly important in fertility modeling this decline varies substantially between regions and over
(Islam et al., 2022; Tzitiridou-Chatzopoulou et al., 2024), time within the same region. Despite this, to the knowledge
enabling machine learning algorithms to uncover hidden of the present researchers, no comprehensive study has
patterns, thus improving the decision-making process. been conducted to study the fertility transition in Ethiopia
using ASFRs or other acceptable detailed indicators,
At the onset of fertility transition, where development primarily due to the lack of reliable ASFR estimates.
is very low, fertility is typically very high and distributed
across various age groups. As development progresses and To address this research gap, this paper proposes
fertility declines, the shape of the fertility curve changes, two main objectives. First, it aims to develop a suitable
with fertility becoming more concentrated in the 20 – 24 methodology for creating an MFT for Ethiopia, which
or 25 – 29 age groups. Thus, observing age-specific fertility can provide plausible estimates of ASFRs for different
curves over time provides a comprehensive view of the regions/provinces of Ethiopia over time from the given
fertility transition in any nation or population. Factors TFR. Second, it seeks to understand the fertility transition
such as increased age of marriage, higher use of modern in regions in Ethiopia from 2000 to 2016 using ASFR
contraceptives, childlessness, and other behavioral estimates obtained from the MFT.
changes contribute to shifts in fertility and the transition The primary goal is to construct an MFT for Ethiopia
process (Mitra, 1967; United Nations, 1963). Therefore, that takes TFR as input and outputs plausible ASFR
examining the ASFR curves of a nation or any region over estimates and to study the fertility transition in Ethiopia
the transition period facilitates a thorough study of fertility and its regions over the past two decades using these ASFRs.

Volume 11 Issue 3 (2025) 55 https://doi.org/10.36922/ijps.4086