Page 231 - GHES-3-1

Global Health Economics and
Sustainability
Fertility models using Nepalese and Malaysian data

teen years in the year 2016 (NDHS, 2017). This number was Various models for describing and predicting fertility
reduced to approximately 14% in the year 2021. Specifically, behavior have been proposed. Two primary types of
10% had had at least one live birth, 2% had experienced fertility models exist deterministic and stochastic models.
pregnancy loss, and 4% were still pregnant (NDHS, 2022). Deterministic models assume that fertility is a fixed
Adverse health effects on mothers and newborns are likely outcome determined by known factors. Meanwhile,
to result from pregnancy and childbirth in the teen years. stochastic models allow for random variations in fertility
Teenage mothers face challenges that prevent them from outcomes. The selection of model type depends on the
completing their education and increase the risk of illness as specific research question under consideration. The
well as child and mother fatalities. Further, young adolescent history of modeling fertility curves dates back to the
pregnancies have been associated with an increased 1940s. The Hadwiger function (Hadwiger, 1940) has three
likelihood of underweight children (Santos et al., 2014). parameters and is helpful for bimodal ASFRs. Gilje &
Teenage mothers are more likely to experience anemia, Yntema (1971) proposed a shifted Hadwiger function by
postpartum depression, and infections, as well as a high introducing a fourth threshold parameter. Hoem et al.
risk of surgical birth and improper beginning of lactation (1981) proposed a probabilistic aspect for investigating
(Jeha et al., 2015). In addition, according to a report by the fertility; they articulated numerous types of probability
World Health Organization (WHO, 2018), pregnancy during density functions (PDFs) using the Hadwiger, Gamma,
adolescence can lead to an increased risk of low weight, Beta, Coal-Trussel, Brass, and Gompertz functions.
preterm birth, and severe health complications for newborns, Furthermore, they employed a deterministic form of the
including neonatal sepsis. These outcomes underscore the fertility curve based on regression spline and polynomial
health challenges adolescent mothers and their infants functions. Chandola et al. (1999; 2002) investigated and
may face, which can have long-term implications for child applied a two-component combination model of Hadwiger
development and survival. in this regard. Peristera & Kostaki (2007) investigated and
applied the normal mixture model to fertility and claimed
Fertility pattern analysis and modeling to smooth that the model is flexible in capturing various fertility data.
ASFRs form a well-established research field globally, Mazzuco & Scarpa (2011) proposed a flexible generalized
possibly representing an emerging research area of inquiry skew-normal (FGSN) distribution. Further, in this context,
within Asian contexts. In developed countries, ASFRs the skew-logistic distribution has been used by Asili et al.
typically exhibit a bimodal skewed fertility curve, whereas (2014) and Mishra et al. (2017); meanwhile, Gaire and
in developing nations, they usually display a unimodal Aryal (2015) employed an inverse Gaussian distribution.
right-skewed fertility curve. This distinction highlights In addition, non-parametric models have been proposed.
the varied demographic landscapes across regions. For For example, Schmertmann (2003) proposed a piece-wise
decades, demographers worldwide have been engaged quadratic spline function. An adjusted error model was
in the mathematical modeling fertility curves. In Nepal, used by Gayawan et al. (2010). Support vector mechanics
research had largely concentrated on identifying specific – a non-parametric technique – was used by Kostaki et al.
factors that influence women’s reproductive behavior and (2009). Different forms of polynomial models have been
fertility patterns, while little attention had been given to used by Islam & Ali (2004), Islam (2009; 2011), Singh et al.
advanced modeling approaches of demographic variables. (2015), and Gaire et al. (2022).
Various researchers have employed smoothing or The mathematical models used for ASFR modeling
statistical graduation techniques to determine the actual are generally divided into deterministic and stochastic
patterns of ASFRs and estimate fertility parameters. categories, or alternatively into parametric and non-
Mathematical modeling is creating an accurate parametric models. Effectively modeling diverse fertility
mathematical representation of a real-world scenario for patterns worldwide is critical, especially in regions with
making predictions or gaining insight into a given subject high fertility rates as well as those approaching or below
(Salomon & Murray, 2001). These techniques provide replacement levels. These models effectively describe the
precise descriptions of the natural shape of ASFR patterns ASFRs in developed areas, where fertility generally as
and aid population projection. Historically, demographers correspond to lower levels. Nevertheless, fertility remains
have employed parametric models, with the parameter high in many countries across Africa, parts of Latin
being the involved mother’s age at childbirth. Mixture America, and region in Asia. Therefore, researchers are
probability models have been proposed to fit bimodal- working or develop flexible, adaptable models capable of
shaped ASFRs. Subsequently, standard non-parametric capturing these diverse fertility patterns, particularly to
techniques, such as kernels, splines, and support vector identify the differentials and determinants. Developing
machines, have been used to smooth ASFRs. countries such as Nepal have a unimodal, right-skewed

Volume 3 Issue 1 (2025) 223 https://doi.org/10.36922/ghes.4219