Globally, fertility is declining but continues to stall in sub-Saharan Africa (Kebede et al., 2019; Schoumaker, 2019). The 2017 Ghana Maternal Health Survey (GMHS) reports that total fertility rate is 3.9 children per woman. Fertility preference of women has been identified as one of the key determinants of actual fertility (Bongaarts & Casterline, 2012; Casterline, 2017; Cleland et al., 2020; Mbacké, 2017). The ability of individuals to make informed decisions on their sexual and reproductive rights is reinforced in global goals. The 2030 Agenda for Sustainable Development Goals (SDG) focuses on making reproductive health and rights a specific aim as stated in SDG 5.6. As the report stated, ‘the way forward is the full realization of reproductive rights, for every individual and couple by dismantling all the barriers whether economic, social or institutional that inhibit free and informed choice’ (UNFPA, 2018; pg. 5).
Studies have established that fertility preference is not stable and changes due to prevailing circumstances (Liefbroer, 2009; Trinitapoli & Yeatman, 2018). Furthermore, studies conducted in sub-Saharan Africa (SSA) including Ghana have identified a number of factors as predictors of high fertility preference (Ahinkorah et al., 2020; Kodzi et al., 2010; Odusina et al., 2020; Van Lith et al., 2013). Findings from these studies suggest positive relationship between fertility preference and actual fertility. Günther and Harttgen (2016) studying developing countries found that women in sub-Saharan Africa are unable to translate fertility preferences into actual fertility, and consequently leading to overachieved fertility. A woman’s inability to achieve her fertility desire could result in two outcomes: underachieved or overachieved fertility. The situation where a woman’s fertility preference is more than her actual fertility is termed as underachieved. On the other hand, if the fertility preference is lower than her actual fertility, it is described as overachieved fertility. The situation where a woman’s preferred fertility is the same as her actual fertility is, however, termed as achieved fertility. Several factors account for the difference between a woman’s preferred fertility and her actual fertility and the consistency of these factors in explaining this gap is the focus of this paper.
Factors accounting for fertility gap
The influence of childhood mortality on fertility behaviour could be explained from the insurance and replacement effect perspective. When childhood mortality is high, couples increase their target family size to ensure that their minimum number of preferred children survive into adulthood (Poppel et al., 2012; Reher et al., 2017). Regarding the replacement effect, studies have found that fertility behaviour could be a sequential decision-making model, where couples may work towards replacing a child who has just died (Gyimah & Ferrnando, 2004; Wolpin, 1998). For instance, Gyimah and Fernando (2004) found that childhood mortality experience has long-term implications on fertility beyond short-term physiological effects in that the death of the first child in particular is associated with the risk of higher order births. We hypothesize that women who have had an experience of child loss are less likely to attain achieved fertility compared to overachieved fertility.
Partner characteristics have also been identified as important factors that predict fertility behaviour in sub-Saharan Africa. Fertility preference is much dependent on the social context. The outcome of a reproductive decision may depend on the social context that favours one reproductive choice or another (Voas, 2003). Within a patriarchal society, husbands have power in making reproductive decisions for their spouses. Therefore, the fertility preference of a man could determine the woman’s actual fertility. In Ghana, a study by DeRose and Ezeh (2005) revealed that reproductive decision-making is determined by men’s characteristics. Men with secondary and higher education were more likely to have a lower fertility preference compared with their counterparts with no education. In a similar study by Dodoo and Landewijk (1996), it was reported that women’s desire to limit fertility is not associated with women having greater control over their reproductive decisions. Their study further revealed through a focus group discussion that the highly educated young women with high fertility preference are not able to achieve their preferred fertility without the approval of their husbands. However, other studies contend that the influence of men in taking reproductive decisions decreases with increasing years in marriage. Bankole (1995) argues that among the Yoruba in Nigeria, husbands take primary responsibility for fertility decisions at early stages of marriage, but their dominance reduces at the later stage of marriage when several children are born. In SSA, a couple’s fertility preference is usually not the same (Muhoza et al., 2014). Having agreement on preferred fertility indicates spousal communication between the couple (Hinson, 2015; Kodzi et al., 2012). This could be that men in the sub-region are pronatalist and may have fertility preferences higher than women. Hence, women having more children than they preferred could be that their husbands preferred the extra number of children they had. Therefore, the study hypothesizes that women are more likely to attain achieved fertility when their partner’s fertility preference is lower.
Female education assists in achieving planned number of births through the pathways of knowledge and access to contraception. Besides, women with female education are characterized with greater autonomy, reduced dependence on sons for social status and old age security (Dreze & Murthi, 2001). Nitsche and Hayford (2020) in their study found that in the United States underachieved fertility is common among women with higher level of education. This is because most of the educated women delay their age of motherhood. Similarly, in a study using data from Demographic and Health Surveys (DHS) in 34 sub-Saharan African (SSA) countries, Kebede et al. (2021) found that desired family size declined with improvements in educational status of women. Hence, it is expected that higher educational level of women will reflect in the difference between women’s fertility preference and actual fertility. Women with higher level of education are expected to attain achieved fertility.
Fertility differences of ethnic groups have been explained from different perspectives. First, it has been explained from the socioeconomic status perspective. For instance, women belonging to ethnic groups with high fertility are found to have low socioeconomic status (Akonor & Biney, 2021; Gyimah, 2002a, 2002b). The other explanation is that cultural norms of ethnic groups regarding fertility may influence fertility. That is, women belonging to ethnic groups that place much value on higher number of children born are more likely to have high fertility. Hence, cultural norms of some ethnic groups encourage high fertility, especially, in sub-Saharan Africa.
The age group 45–49 years is typically a period when a woman is ending her reproductive life and fertility is low or virtually non-existent. For instance, in Ghana, age specific fertility rates per 1000 women for 3 years preceding the survey among women aged 45–49 years declined from 61 in 1988 to 16 in 2017 (Ghana Statistical Service et al., 2015; Ghana Statistical Service and Macro Inc., 1998). Nevertheless, some studies examining the relationship between a woman’s preferred and actual fertility used a sample of women aged 15–49 years. This does not give a deeper understanding of the phenomenon, since fertility preference of a woman is subject to change especially at the early years of a woman’s reproductive life. To date, there is very little research examining the relationship between a woman’s preferred fertility and actual fertility using women aged 45–49 years (Canning et al., 2013; Carvalho et al., 2016; Casterline & Han, 2017; Ibisomi et al., 2011). Comprehensive understanding of the relationship among women aged 45–49 is critical to advancement of knowledge about the demographic transition as observed in SSA including Ghana.
Studies have recorded that the pace of fertility decline in SSA is inconsistent with its rate of social and economic development (Bongaarts, 2017). This defies the conventional demographic transition theory. Rosero-Bixby and Casterline (1993) explained that fertility transitions are characterized by four main stages. These are pre-transition, onset of decline, mid-transition and post-transition levels of fertility. Agyei-Mensah (2006) has argued that Ghana is at the mid-transition stage characterized with slow or stalled fertility (Agyei-Mensah, 2006). According to Agyei-Mensah (2006) the attainment of fertility preference at this stage is very critical to fertility decline. One of the questions that remain unanswered by demographers is whether the difference between fertility preference and actual fertility at this stage of the transition could be one of the causes of stalling fertility in SSA.
With SSA characterized with overachieved fertility, the rising trend of underachieved and achieved fertility needs to be investigated to better understand the fertility behaviour of couples in sub-Saharan Africa (Channon & Harper, 2019). While overachieved fertility has been associated with child loss, low education and rural places of residence, achieved fertility has been associated with partners’ low fertility desire and differences in ethnicity (Channon & Harper, 2019; Yeboah et al., 2021). Though the decline of overachieved fertility may be an indicator of fertility decline, a woman’s ability to attain her fertility desire may not be a panacea to the slow pace of fertility transition in SSA. Rather, policy makers should be concerned with the nominal value of her fertility desires that was achieved.
There has been a plethora of studies to explain why most women in sub-Saharan Africa exceed their fertility preference. Little evidence, however, exists on the predictors of achieved and underachieved fertility. The purpose of this analysis is to assess the trends and identify consistent determinants of underachieved and achieved fertility relative to overachieved fertility in Ghana from 2003 to 2014 among women who have completed their fertility. The most consistent determinants will be identified to serve as intervention areas for the Ghana Health Service, the National Population Council of Ghana and other relevant partners to inform their programming activities towards further fertility reduction in Ghana.
Yeboah et al. (2021) have shown that women are unable to attain their fertility desires due to factors, such as marital experiences, child loss experience and unmet need for family planning. The paper investigated the predictors of fertility gap using pooled data sets from GDHS for over a period of 20 years and controlled for survey years. With this study, we aim to extend and deepen the discussion by Yeboah et al. (2021) by examining the changes and consistency in factors predicting the fertility gap over the 10-year period. In addition, within the 10-year period, Ghana has experienced a stall in fertility compared to the 20-year period used in the paper by Yeboah et al. (2021).
Source of data and methods
The data for analysis in this study consisted of women who were currently married or living/cohabiting with a partner in the 2003, 2008, and 2014 Ghana Demographic and Health Survey (GDHS) data set who were of age 45–49 years. These categories of women were selected for analysis, because they were asked questions concerning the fertility preference of their partners. The sample for analysis was also restricted to women with at least one live birth, who gave numeric responses to the question on ideal number of children and at the same time responded that they ‘do not want more children’ to the question on desire for more children. The study was restricted to women with at least one live birth, because the study controlled for child loss experience of these women. Besides, child mortality was computed using birth history of women in the study. Women with numeric responses were selected, because the inclusion of the non-numeric responses would have been difficult to determine how their actual fertility could differ from their fertility preference by quantification. The reason for including those who ‘do not want more children’ is because such women have been found to be usually less likely to have a child than those who want more children, especially in sub-Saharan Africa (Cleland et al., 2020). Those who were infecund or sterilized were excluded from the analysis, because infecundity is random concerning fertility preference (Casterline & Han, 2017). The weighted sample sizes that were used for the analysis were: 304, 252 and 474, respectively, for 2003, 2008, and 2014.
Study outcome variables
The difference between fertility preference and actual fertility (fertility gap) was the outcome variable of this study. Fertility preference variable was derived from the ideal family size question in the DHS with emphasis on women who gave numeric responses. Actual fertility variable was derived from the number of children ever born question in the DHS. Women with ideal family size higher than actual fertility were recoded as having ‘underachieved fertility’, while those with ideal family size lower than actual fertility were recoded as having ‘overachieved fertility’. Women with ideal family size as the same as actual fertility were, however, recoded as recording ‘achieved fertility’. Hence, the difference between preferred and actual fertility yielded three results: underachieved, achieved and overachieved fertility.
Predictor variables
The following variables were selected based on the available literature and their relevance to the study: child loss experience, age at first birth, number of marital unions, ever use of modern contraceptives, educational attainment (in years), partner’s education, couple’s fertility preference, place of residence, religion and ethnicity. These variables were categorized as follows: child loss experience [‘All survived’, ‘one or more infant deaths’, ‘one or more child deaths’ and ‘both infant and child deaths’], couple’s fertility preference [‘Both want same’, ‘partner wants more’, ‘partner wants fewer’, and ‘don’t know’], age at first birth [under 18 years; 18–20 years; and 21 years and above]; number of marital unions [once, and more than once]; ever-use of modern contraceptives [yes and no]; educational attainment [in years]; place of residence [urban and rural]; religion [Catholic; Other Christian; Islam; or Other] and ethnicity [Akan; Ga-Dangme; Ewe; Mole-Dagbani and Other].
Statistical analysis
Descriptive techniques of analysis including the use of percentages were used to describe the differences among the women by fertility gap. At the multivariate level, the multinomial logistic regression model was used to examine the net influence of possible individual socio-demographic and spousal characteristics on underachieved and achieved fertility compared to overachieved fertility among currently married women aged 45–49 years who have completed fertility. Women with overachieved fertility were used as the reference category, because it was assumed that most women in sub-Saharan Africa end their reproductive life by attaining overachieved fertility. This study, once again seeks to understand what factors consistently influence women to deviate from overachieved fertility to realize either underachieved or achieved fertility over the past 10 years. The model predicted underachieved and achieved fertility relative to overachieved fertility among women who have completed their fertility. Additional bivariate analysis was performed to explore differences in fertility behaviour and ethnic group affiliation of women. Thus, association between ethnic group and fertility behaviour (fertility preference and actual fertility) was tested using compare means and one-way analysis of variance.