The Interplay of Mortality,Economics,and Female Empowerment in Fertility Transformations

Introduction

Theoretical explanations for the fertility patterns associated with the demographic transition have attributed primary causal influence to one of three factors: (a) economic development (Caldwell, 1976; Harris & Ross, 1987), (b) decreases in population mortality rates (Chisholm et al., 1993; Coale, 1984), and (c) transmission of norms associated with female empowerment (Cleland & Wilson, 1987). Studies examining these factors, considered collectively, indicate the relative contribution of each in explaining how fertility transformations vary across time periods and both within and across scales of analysis (i.e., local, regional, and national levels; Mason, 1997; Matthiessen & McCann, 1978; Sanderson & Dubrow, 2000; Shenk, Towner, Kress, & Alam, 2013; Van de Walle, 1986). Identification of the factors (e.g., sociocultural, economic, political) ultimately responsible for these spatiotemporal variations will be aided through models incorporating pooled cross-cultural and time-series data. Analysis of such data facilitates the development of a “unified” theory of demography by determining the relationship between fertility and the three factors across differing levels (e.g., non-linear at high mortality levels) as well as the interactions across these levels.

To examine the interplay of economics, population mortality, and female empowerment in transforming global fertility patterns, the current article incorporates data from 167 countries across a 40-year time period (1970-2010). A fixed effects approach is used to model the within-country contributions of each factor on reproductive onset (adolescent fertility) and output (total fertility). The model tests for the presence of non-linear effects and interactions between the three factors. In general, results indicate each variable displays a non-linear association with fertility, while the effect of economics and female empowerment varies across levels of population mortality.

The Three Influences

Method

Data

Data were collected from 167 counties from the World Bank (http://data.worldbank.org/), representing decadal time spans from 1970 to 2010 (see the appendix for summary statistics and correlation matrix). Dependent variables were adolescent fertility and total fertility rates. Total fertility rate (tfr) is the expected reproductive output, given current age-specific fertility rates. Adolescent fertility (asf) is the birth rate for females aged 15 to 19 years per 1,000 individuals. Population mortality was represented by the rate of infant mortality (infmort), that is, the number of deaths per 1,000 births. Given the presence of significant skew and kurtosis, infant mortality rates were square-root transformed. Economics was represented by changes in a nation’s GDP per capita (gdp) and was log transformed. Norms of fertility and female empowerment were represented by the gender parity measures of female-to-male ratio in secondary (fmsecond) and tertiary enrollment (fmtert). Different education levels were used as secondary enrollment is representative of female empowerment changes in developing countries, while tertiary enrollment is a better indicator in more developed countries, especially historically. If data were missing for a particular time period (e.g., 1970), the point was filled in with data spanning a ±3-year time period (i.e., 1967-1973). Table 1 provides summary statistics across the five time periods. Table 2 provides a correlation matrix.

OPEN IN VIEWER

Table 1.

Mean and Observations Across Five Time Periods.

	1970		1980		1990		2000		2010
	M	Count	M	Count	M	Count	M	Count	M	Count
Total fertility	5.19	179	4.64	179	4.03	179	3.34	179	2.96	181
Adolescent fertility	95.73	180	87.22	180	75.86	180	63.23	180	52.18	180
Infant mortality	86.61	138	66.03	173	50.03	181	40.20	181	28.88	181
GDP per capita	882.97	119	4,187.09	135	5,073.84	169	6,290.19	177	12,096.41	175
GRT	4.70	129	9.35	129	15.03	138	25.66	152	39.95	145
GRS	69.92	135	80.77	137	89.87	149	94.27	160	96.01	160

Note. GRT = gender ratio tertiary; GRS = gender ratio secondary.

OPEN IN VIEWER

Table 2.

Correlation Matrix.

	Total fertility	Adolescent fertility	Infant mortality	GDP per capita	GRT	GRS
Total fertility	—
Adolescent fertility	.773*	—
Infant mortality	.838*	.740*	—
GDP per capita	−.450*	−.461*	−.470*	—
GRT	−.684*	−.571*	−.765*	.280*	—
GRS	−.662*	−.562*	−.620*	.583*	.440*	—

Note. GRT = gender ratio tertiary; GRS = gender ratio secondary.

*

p < .001.

Results

Final model results are provided in Table 3. Regression coefficients are interpreted as within-country changes in the parameter are associated with within-country changes in fertility. Models 1 and 3 (main and quadratic effects) indicate that infant mortality was positively associated with adolescent and total fertility although this association decreased at higher levels of mortality. Confidence intervals associated with this relationship (see Figure 1), however, indicate that precision substantially decreases at both high and low mortality rates. GDP per capita was negatively associated with both fertility rates although this effect decreased at higher capita levels. Gender parity in tertiary education was negatively associated with total fertility but not significantly related to adolescent fertility. In contrast, gender parity in secondary education was significantly related to decreases in adolescent fertility but not total fertility.

OPEN IN VIEWER

Table 3.

Model Results.

	Total fertility		Adolescent fertility
	M1	M2	M3	M4
Infant mortality	0.83**	2.14**	11.08**	59.66**
Infant mortality2	−0.03**	−0.06**	−0.48**	−1.58**
GDP per capita	−1.19**	1.63**	−12.11*	43.31**
GDP per capita2	0.09*		1.14**
GRS	−0.01	0.06**	−0.96**	2.71**
GRS2	0.0001		0.004**
GRT	−0.01*	0.006*	−0.10	−0.14
GRT2	0.0001		−0.0005
Infant mortality × GDP		−0.11**		−2.62**
Infant mortality × GRS		−0.001		−0.18**
Infant mortality × GRT2		−0.001*		0.02
GDP × GRS		−0.008**		−0.25**
1980	0.13	−0.15	−7.75**	−7.97**
1990	−0.08	−0.49**	−16.01**	−15.71**
2000	−0.21	−0.76**	−22.01**	−20.60**
2010	−0.08	−0.86**	−30.31**	−28.21**
Constant	3.68**	−12.67**	110.35**	−398.91**
R 2	.84	.93	.83	.92
N	575	575	575	575
Countries	167	167	167	167

Note. See “Method” section for description of variables. GRT = gender ratio tertiary; GRS = gender ratio secondary.

OPEN IN VIEWER

Figure 1.

Within-country impacts of infant mortality on total fertility and adolescent fertility.

The addition of interaction effects (see Models 2 and 4) rendered several quadratic effects non-significant and these quadratic effects were removed. Interaction effects were graphed using three different levels of the modifier variables (i.e., mean & ± 1 standard deviation; see Figure 2). In general, the effects of economic and gender parity on fertility varied across mortality levels. Gender parity in secondary education was negatively associated with total and adolescent fertility but only in countries with medium and high levels of mortality. Likewise, total fertility was only effected by gender parity in tertiary education in medium and high mortality levels. For adolescent fertility, however, the negative relationship with gender parity in tertiary education was only significant within low mortality countries. GDP exhibited a positive relationship with total fertility rates within low and medium but not high mortality countries. For adolescent fertility, GDP has a negative relationship at high mortality levels, a positive relationship at low mortality levels, and a non-significant relationship within medium mortality levels. A significant interaction between GDP and gender parity in secondary education (not pictured) indicated that wealth and secondary education display a synergistic effect on fertility, although only at medium and high wealth countries.

OPEN IN VIEWER

Figure 2.

Effect of gender parity and GDP on within-country total fertility and adolescent fertility rates at low, medium and high mortality.

Discussion

Using a fixed effects approach, the current study examined the association between within-country changes in economics, mortality, and female empowerment and the onset and output of female reproduction within 167 countries across a 40-year period (i.e., 1970-2010). In general, fertility rates were positively associated with mortality rates and negatively associated with economic development and advances in female empowerment. However, this relationship was not linear across all levels of the three factors. Higher levels of population mortality and economic development were associated with decreasing impacts on fertility rates. These results suggest the presence of threshold effects. For example, it may be that mortality must decline to a certain level before resulting in statistically noticeable changes in reproductive behavior. Indeed, the fertility decreases observed in countries with extremely high mortality environments (see Figure 1), beyond issues of small sample size, may not reflect optimal life history strategies but instead very low infant survivorship. This interpretation would be consistent with previous studies finding a negative or no relationship between mortality and reproduction in sub-Saharan Africa and countries low on the Human Development Index (Cantrelle et al., 1978; Low et al., 2008)

The association between the three factors and within-country fertility rates across the 40-year period also varied across levels of the other factors. Decreasing rates of gender parity in secondary and tertiary education were associated with the largest decrease in total fertility rates in medium and high mortality countries. In contrast, gender parity decreases in low mortality countries did not result in significantly different total fertility rates. Thus, female empowerment norms, at least those related to female education, have the greatest impact on fertility in high mortality environments. These results seem to contradict suggestions that education and economic factors begin to take precedence in fertility decisions only when mortality rates fall (Kaplan, 1994; Shenk et al., 2013). Alternatively, a non-significant relationship in low mortality countries may occur as gender disparities in education within these countries have exhibited relatively smaller changes over the last 40-year period (i.e., no variability). Indeed, the threshold may be shifting so that higher levels of female empowerment (e.g., women as the “breadwinners” of the household) need to be attained before resulting in statistically noticeable impacts on reproduction. This interpretation would be consistent with Model 3 showing that as gender parity in secondary education increases the negative association adolescent fertility decreases. Future studies should incorporate indicators that better represent female empowerment in developing countries, such as the relative economic contribution of a woman to the household controlling for other female empowerment indicators.

An unexpected finding, in light of theoretical expectations and previous empirical work, was the synergistic interaction between economic growth and fertility rates (see Figure 2). GDP increases resulted in the largest total fertility increases within low mortality countries. These findings are even in contradiction with Model 1 in the current study that documented a negative relationship between GDP and total fertility. ² The trend is likely a product of nations that have experienced large growth in GDP over the last 40 years but remain within Stage 3 of the demographic transition, where mortality rates have decreased but fertility rates remain high, including several Arab states. This trend may provide support for ideational theories if the relatively lower rates of gender-equality observed in Arab societies pressurize women toward earlier and more frequent reproduction (Norris & Inglehart, 2001). Combined with economic growth that allows for increases in parental investment, such pressures may lead to dramatic increases in fertility. This possible synergistic interaction between economic growth and cultural norms in some low mortality countries demonstrates why fertility models should consider all three proposed influences in concert to provide a more nuanced account of global fertility patterns.

The current study has several limitations. First, lacking direct measures of female empowerment/fertility norms, the current models cannot directly test ideational theories. If indirect indicators, (e.g., gender parity indices) are not correlated with these norms, then model results should not be considered a proper gauge of association. In fact, a study of nine Latin American countries found that while female education was negatively related to total fertility, most females shared the desire for a small family size regardless of education level. However, better educated women had a greater breadth of knowledge (e.g., contraception methods) and higher socioeconomic status (Martin & Juarez, 1995). Thus, while low fertility norms are necessary, they do not seem sufficient when lacking access to key resources, such as education. Gender disparity in education then, although not a direct indicator, is likely a reliable indicator of the propensity of norms to affect fertility. Again, such dependencies highlight the importance of including all factors, as well as interactions, in models of fertility behavior. A further limitation of the current study is that effects were modeled as contemporaneous over the 40-year period and so must be interpreted as associations. A stronger case for causation could have been made through the specification of lagged effects (Mason, 1997). However, current data availability ensures that the specification of lagged effects produced a substantial decrease in sample size. ³ A further complication is that some influences may exhibit shorter/longer lagged effects than others (e.g., economic versus mortality changes; Mason, 1997). Future studies should examine such temporal differences as they likely vary across nations with differences in cultural, economic, and political contexts.

Conclusion

The current study applied a fixed effects approach to examine how the onset and output of female reproduction was affected by economics, population mortality, and female empowerment within 167 countries across a 40-year period (1970-2010). In general, results indicate that the three factors affected fertility in theoretically consistent ways. However, impacts of the influences were not linear and interacted with levels of the other influences, especially mortality, in the transformation of fertility patterns. These deviations provide a point of departure for examining how cultural differences across and within nations (e.g., short-term vs. long-term orientation) affect the relationship between the three proposed influences and fertility patterns. The development of a “unified” theory of demographic transitions will be supported by cross-cultural examinations that can establish whether the fine-grained results of local and regional analyses reflect global patterns and, if not, begin to identify the local/regional influences (e.g., cultural, political, historical) that give rise to these variations.