The Impact of Female Labor Force Participation on Enhancing Human Development Throughout the Stages of Global Demographic Dividends

The Impact of Female Labor Force Participation on Enhancing Human Development Throughout the Stages of Global Demographic Dividends

Chulan Lasantha Kukule Nawarathna

Department of Social Statistics, University of Sri Jayewardenepura, Sri Lanka.

DOI: https://dx.doi.org/10.47772/IJRISS.2025.90400045

Received: 23 March 2025; Accepted: 27 March 2025; Published: 26 April 2025

ABSTRACT

The low female labor force participation rate (FLFPR) must increase to empower women and enhance the demographic dividend (DD) to promote human development. The United Nations Human Development Index serves as the fundamental measurement tool for assessing progress in human development, evaluating the three components of healthcare delivery, educational standards, and economic productivity. Research investigating the global impacts of the female labor force participation rate on the Human Development Index reveals its influence at different demographic dividend stages, from Pre-DD to Early-DD, Late-DD, and Post-DD, from 1990 to 2019. The female labor force participation rate exhibits a significant positive correlation with the Human Development Index at all phases of the demographic dividend period, according to results obtained from Driscoll-Kraay standard error regression and Dumitrescu–Hurlin causality testing. The female labor force participation rate shows an inverse U-shaped relationship with the Human Development Index in both the Pre-DD and Post-DD periods, yet demonstrates a U-shaped pattern in Early-DD and Late-DD across all regions and globally. The female labor force participation rate and the Human Development Index remain directly connected during every demographic dividend period. Population growth and per capita GDP have been confirmed to enhance the global Human Development Index while benefiting all phases of demographic dividend development. The observed evidence indicates that implementing appropriate policies to optimize female labor force participation is essential to access all advantages from demographic dividends. Through gender equity policies, policymakers can enhance human development in every aspect by strategically managing populations alongside economic improvements for women.

Key Words: Human Development Index, Female labor force participation rate, Demographic dividends, Panel data analysis. Women empowerment.

INTRODUCTION

Human development is a crucial factor in determining a country’s level of development (Arisman, 2018). Based on three components: life expectancy, expected years of schooling, and per-capita gross national income, the Human Development Index (HDI) was introduced by the United Nations Development Program (UNDP) in 1990 as a new method for evaluating human development that was based on Amartya Sen’s “capabilities approach” (UNDP, 2016). HDI is frequently used to calculate the level of human development in all countries (I. Ivanova, 1999). Although the HDI has been criticized for not considering gender inequality and ecological concerns, and it does not predict future development levels (Sagar & Najam, 1998), it is the most popular composite indicator for evaluating and comparing a country’s socioeconomic development globally (Dasic et al., 2020).

The world’s population has quadrupled since the mid-20th century and is expected to reach 11 billion by 2100 (John Wilmoth,  2022). The “demographic transition” is the change from high fertility and mortality rates in agrarian societies to low rates in urban industrial communities, according to Lee (2003) and R. D. Lee & Mason (2006)During this transition, a “first demographic dividend” occurs, leading to a temporary rise in the labour force growth rate and a decline due to decreased fertility. This ultimately results in a reduction in the labour force growth rate and a stagnation in the increase of per capita income, giving rise to a “second demographic dividend.”

Mostly, countries in the Late and post-demographic dividend (DD) stages have a high level of HDI, and countries in the pre- and Early stages have a low HDI (Figure 1). Also, highlighting the significance of the impact of the size of the working-age population ( the first DD) on economic growth, Ahmed et al. (2016) show that the economic status of countries mostly aligns with their DD stages. Although gender composition is equal, FLFPR is significantly low globally. Effective policy is needed to maximize the DD and encourage FLFP (Lee & Mason (2019). FLFP has the potential to create multiple social and economic benefits. Improvement in FLFPR increases economic growth, reduces gender disparity, and advances women’s societal positions, resulting in better health, higher literacy education levels, voluntary control over population growth, and equitable income and benefit distribution. Also, gender equality can accelerate economic growth in fields associated with the labor market, employment, education, and revenue in countries with high HDI (Gelard & Abdi, 2015).

Furthermore, HDI indicates sustainable human development by combining health, human capital, and economic indices. FLFP may directly and indirectly impact the United Nations Sustainable Development Goals (SDGs) (Foster, 2016; Balakrishnan & Dharmaraj, 2018; Denney, 2015; Taheri et al., 2021). The economic, social, and environmental pillars of sustainable development can all benefit from FLFP (Choudhry & Elhorst, 2018; Ustabaş & Gülsoy, 2017; Appiah, 2018)This study explores how FLFPR affects HDI at different demographic dividend stages and globally, considering the linkage between FLFP, DD stages, and HDI. The findings may help policymakers increase HDI by obtaining the maximum contribution from the FLFP of the countries on their DD stages and globally.

Previous research has identified several factors of HDI (Table 1). Various economic, demographic, and energy-related factors influence HDI at the national and regional levels. This study employs Population, per capita GDP, per capita energy consumption, and male and female labor force participation rate to explain HDI at each dividend stage and globally.

Table 1: Brief Summary of Studies on the HDI as a Dependent Variable.

Created by the author

The study employs several panel econometric methods, including slope homogeneity tests, second-generation unit root tests, Westerlund cointegration tests, Driscoll and Kraay, and Newey-West regressions for long-run estimation, as well as the Dumitrescu-Hurlin Granger non-causality test to identify causal relationships among variables. This study comprises five segments: Segment 2, Model Specification and Data Sources, The Theoretical Framework, Modelling, and Data Collection. Segment 3, The Estimation Strategy, presents a range of econometric techniques in this study. Segment 4, Empirical Results and Discussion, describes the analysis and empirical discourse. Segment 5, Conclusions, offers concluding remarks on practical implications and suggestions for future research.

Model Specification and Data Sources

Theoretical framework

The HDI is the geometric mean of the three main indexes: the Life Expectancy Index (LEI), the Education Index (EI), and the Income Index. Under the above indexes, the following factors used to calculate the HDI include 1) a long and healthy life a population (life expectancy at birth), 2) a decent standard of living (GNI per capita (in PPP adjusted international-$)), 3) access to knowledge (Mean years of schooling & Expected years of schooling). Shah (2016), Wang et al. (2021), Barus et al. (2021), Humaira & Nugraha (2018), Khan et al. (2019), and  Arisman (2018) have shown that factors such as GDP, life expectancy, and literacy rate have a positive impact on the HDI. However, the study by (Wang et al., 2018) shows a negative effect on HDI by GDP in Pakistan. Also, studies by Humaira & Nugraha (2018), Akbar et al. (2021), and  Barus et al. (2021) confirm that life expectancy and expenditure in the health sector improve HDI. According to the findings of Humaira & Nugraha (2018), Haque & Khan (2019), Barus et al. (2021), Haque & Khan (2019), and Madhusudan Ghosh (2018), the impact of educational factors on the HDI is positive. Other than the variables directly related to the HDI’s main components, researchers work to identify the effects of different factors such as the economic behavior of the government, energy, and demographic features.   Tripathi (2019), Shah (2016), and Arisman (2018) show that inflation negatively impacts HDI. Also, the Gini coefficient, research and development expenditure,  trade, FDI, and public debt reduce the HDI according to Shah (2016), Akbar et al. (2021), Khan et al. (2019),  Wang et al. (2018) and Wang et al. (2021) respectively. According to Wang et al. (2018), renewable energy reduces the HDI in Pakistan, but Wang et al. (2021) show that renewable energy increases the HDI in OECD countries.   Ouedraogo (2013) shows that per capita energy consumption reduces HDI, but electricity consumption improves HDI in developed countries. Population and population growth rates reduce HDI, according to Zgheib et al. (2006), Arisman (2018), Akbar et al. (2021), and Tripathi (2019). Wang et al. (2018) and Khan et al. (2019) show that urbanization also reduces the HDI. Considering the findings of those studies, the impact of FLFPR on the HDI is assessed, considering population, gross domestic product, energy usage, and male labour force participation rates as the controlled variables.

Specifications of the Empirical Model

To explore the main effect of FLFPR on HDI in the countries across demographic dividend stages. This study considers the following equation:

                                                                                (1)      

HDI is the human development index, POP is the total midyear population, GDP is the per capita gross domestic production indicator, ENG is per capita energy consumption, FLFPR is the Female Labor Force Participation Rate, and MLFPR is the Male Labor Force Participation Rate.

The nonlinear effect of FMLFPR on human development can be explored by adding the square term of the FLFPR term to equation (1), as shown in equation (2):

                                                                   (2) 

In line with Sinha & Sen (2016) and Tran et al. (2019)We explore how several variables relate to one another within a single multivariate framework. Bekhet & Othman (2017) noted that all the data must be transformed into natural logarithms to lower the likelihood of autocorrelation and heteroscedasticity on that all data convert into natural logarithms. Also, by reducing the sharpness of the data, the log-linear model provides more trustworthy conclusions than the basic model. (Shahbaz, 2013). The empirical model for this study is specified as follows:

In models (3) and (4), β_1, β_2, β_3,β_(4,),β_5 and β_6d reflect the elasticity relations between the independent variable and dependent variables. Every 1% change in LPOP, LGDP, LENG, LFLFPR, LMLFPR, and LFLFPR2 leads to a β_1, β_2, β_3,β_(4,),β_5 or β_6 change in HDI.

The study utilised reliable databases to acquire secondary data (Table 2). The Human Development Index (HDI), which served as the study’s response variable, was sourced from the United Nations Development Programme website. Data on mid-year total population (POP) and per capita Gross Domestic Product (GDP) in constant 2015 US dollars were obtained from World Development Indicators. The BP & Shift (2020) Data Portal supplied data on energy usage (per person, 2020) in kilowatt-hours for the study. The study also gathered data from World Development Indicators for the explanatory variables of the Female Labour Force Participation Rate (FLFPR) and the Male Labour Force Participation Rate (MLFPR).

Table 2: Variables of the Study and Data Sources

Created by the author

Classification of the study panels.

A global classification of nations based on demographic characteristics was created by Ahmed et al. (2016) using the first demographic dividend theory proposed by R. D. Lee & Mason (2006). They are classified as pre, early, late, and post-dividend countries. Based on this, 191 global countries were classified into four stages of demographic dividends in the World Bank’s World Development Indicators database. This international study analyzes such subpopulations depending on the country’s classification indicated above. Figure 1 depicts a visual representation of the classification of world countries based on their demographic dividend stage and HDI.

Figure 1: Demographic dividends and Human Development Index around the world.

Source:

1. Global Monitoring Report 2015/2015, worldbank.org/gmr
2. Reserved from the Max Roser (2014) – “Human Development Index (HDI).” Published online at OurWorldInData.org. Retrieved from: ‘https://ourworldindata.org/human-development-index’ [Online Resource]

According to the aforementioned classification, thirty-seven countries worldwide are in the initial demographic dividend stage, the pre-demographic dividend stage. Early, Late, and Post demographic dividends correspond to the second, third, and fourth demographic dividend stages in 62, 54, and 38 countries, respectively. This panel study selects 118 countries for the Global Panel and 20, 41, 29, and 28 countries for the pre-, Early, Late, and demographic dividend country panels, covering the period from 1990 to 2019 (Appendix A). Tables 3 and 4 present the descriptive statistics of the variables and their correlation matrices in natural logarithms.

Descriptive Statistics of Study Variables

Across four dividend stages, the results reveal an upward trend in LHDI, LGDP, and LENG, while LPOP and LMLFPR display roughly equal central tendencies. Average LFLFPR exhibits a U-shaped trend across the dividend stages, recorded at 60.15%, 47.88%, 54.54%, and 63.39% during the Pre, Early, Late, and Post demographic dividend stages, respectively. The correlation between LPOP and LHDI is positive in the Pre- and Post-dividend stages but harmful in the Early and Late stages. Although LFLFPR has a negative relationship with LHDI in the Pre- and Early-dividend stages, it shows a positive correlation in the Late, Post-dividend, and Global panels, while LGDP and LENG demonstrate a positive association with LHDI. In the Pre, Early, and Late stages, LMLFPR and LHDI have a negative correlation; however, the Post and Global panels reveal a significantly positive correlation. While LMLFPR has a low standard deviation, LHDI exhibits a high standard deviation across all dividend stages. The standard deviation of LFLFPR is high, whereas LGDP and LENG show a lower standard deviation in the Late and post-periods compared to the Pre and Early dividend stages.

Table 3: Descriptive statistics

Authors calculations

Table 4: Pairwise Correlation

Pre-Dividend Panel

Early-dividend Panel

Post-Dividend Panel

Global Panel

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

Estimation strategy

This research will employ several econometric techniques, including panel pretests such as the slope homogeneity test, the cross-sectional dependence (CD) test, the CADF and CIPS unit root tests, and error-correction-based panel cointegration tests. The study applies panel long-run estimation methods: Driscoll and Kraay standard errors for coefficients estimated via pooled OLS and Newey-West standard errors for OLS regression in linear cross-sectional time series models. The Dumitrescu-Hurlin Panel individual causality estimation test will also account for heterogeneity, cross-sectional dependence, and autocorrelation, thereby providing more accurate results for panel data.

Slope Homogeneity Tests

Swamy (1970) developed the methodology to assess the homogeneity of the slope coefficients of the cointegration equation. Hashem Pesaran & Yamagata (2008) improved Swamy’s slope homogeneity test by developing two “delta” test statistics as in equations (5) and (6).

:  and .

                            (5) (6)           

N represents the number of cross-sectional units, S for the Swamy test statistic, and k for the independent variables. The null hypothesis is accepted at a 5% significance level, and the cointegrating coefficients are thought to be homogeneous if the test’s p-value is more significant than 5%. For large and small samples, respectively, ∆ ̃ and ∆ ̃_adj are suitable, with ∆ ̃_adj being the “mean-variance bias adjusted” variant of ∆ ̃ . Therefore, the error must not be autocorrelated to perform the usual delta test (∆ ̃). A Heteroscedasticity and Autocorrelation Consistent (HAC) resilient variant of the slope homogeneity test was created by Blomquist & Westerlund (2013) by loosening the homoscedasticity and serial independent assumptions of Hashem Pesaran & Yamagata (2008) as in equations (7) and (8).

 and :

                                   (7)          (8)                                                                                  Cross-sectional dependence tests

Due to the interdependence of the countries on a regional and international scale, cross-sectional dependency frequently occurs in panel data. According to Peter C. Phillips and Donggyu Sul (2003), studies that do not consider cross-sectional dependence would yield inconsistent and distorted findings. As a result, it is essential to examine the cross-sectional dependency in the panel data. This study uses three tests to determine the cross-sectional interdependence of the selected variables. N. Bailey, G. Kapetanios (2015) along with Bailey et al. (2019), Chudik & Pesaran (2015), and Pesaran (2004)  CD tests will likely be employed to determine if cross-sectional dependency exists in the residuals of the estimable model.

The succeeding equation of the Bailey, Kapetanios, and Pesaran Cross-Sectional Dependence test (equation (9)) is used to detect the study variables:

                                                                                                                          (9)

Also, the succeeding equation (10) of the CD test is used to undercover the cross-sectional dependence proposed by Pesaran (2004):

                                                                                                  (10)

Where N represents the sample size, T indicates the period and  It shows the estimate of the cross-sectional correlation of errors in countries i and j.

Panel unit root tests

The first-generation unit root findings are inoperable in cross-sectional dependency, according to Dogan & Seker (2016). This study uses the augmented cross-sectional IPS (CIPS) and augmented cross-sectional ADF (CADF) approaches to determine the variables’ stationarity properties. To verify the unit root, Pesaran (2007) Proposed the following equation (11) of the IPS cross-section augmented version:

                                                                                  (11)

Where stands for the difference operator,  displays the variable under analysis, α is a personal intercept, T stands for the time trend in the data, and it is the error term. The duration of the lag is determined using the Schwarz Information Criterion (SIC) method. The null hypothesis for both tests posits that none of the individuals are stationary within the data. In contrast, the alternative hypothesis asserts that at least one individual is stationary within the time series panel data.

Panel Cointegration Test

The Westerlund cointegration test is used in this study to look at the long-run equilibrium between the model variables. Westerlund (2007) using structural dynamics, four actual panel cointegration tests are suggested without the traditional factor constraints. The significance of the error correction component is investigated using a constrained panel error correction model, and the bootstrapped p-values are robust to cross-sectional dependence.

For the complete panel (Gt and Ga), the Westerlund cointegration test uses two tests to investigate the alternative hypothesis of cointegration. The two additional tests, however (Pt and Pa), assess the possibility that at least one cross-sectional unit is cointegrated. Group statistics covers the first two tests, whereas panel statistics covers the latter. An average statistics analysis follows the computation of group-mean statistics and the independent evaluation of the error-correction constants for each cross-sectional unit. The null hypothesis for this technique is “no error correction.” If the null hypothesis is disproved, there is evidence that the equation’s variables are cointegrating. Westerlund considers the subsequent error-correcting model as in equation (12).

                                      (12)

Where I denote the cross-sections, t denotes observations, and it represents the deterministic components, dt computes the convergence speed to the equilibrium state after an unanticipated shock.

Panel long-run estimation method

It is necessary to do an accurate and trustworthy estimate since autocorrelation, heteroscedasticity, and cross-sectional dependence may prohibit the standard fixed effect model from producing unbiased and effective results. According to Wang et al. (2021), cross-sectional dependence renders the estimated results from conventional techniques like FMOLS and DOLS neither accurate nor trustworthy. Therefore, we use Driscoll & Kraay’s (1998) standard error approach to estimate long-run coefficients in this study, similar to the research of Wang et al. (2021), Kongbuamai et al. (2020), Baloch et al. (2019), Hashemizadeh et al. (2021), and Rahman & Alam (2022).

This comprehensive approach considers the estimated model’s autocorrelation, heteroscedasticity, and cross-sectional dependency issues. Driscoll & Kraay’s (1998) standard error technique has several advantages over many other approaches, including the ability to be used with unbalanced panel data, the ability to account for missing values in the dataset, the fact that it is a non-parametric procedure with flexible features and a more significant time dimension, and, most importantly, the ability to accurately correct for heteroscedasticity, autocorrelation, and cross-sectional dependence issues (Hoechle (2007); Rahman & Alam (2022); Wang et al. (2021); Kongbuamai et al. (2020); Baloch et al. (2019)).

After estimating the findings using Driscoll & Kraay’s (1998) standard error method, the robustness of the results needs to be assessed using another well-known two-panel standard error estimating strategy. Regression is performed following the Wang et al. (2021) method’s Newey-West standard errors (Newey & West, 2010)Additionally, these models effectively and satisfactorily address the issues of autocorrelation, heteroscedasticity, and cross-sectional dependence.

Dumitrescu and Hurlin panel causality test

The findings of the Driscoll-Kraay standard error estimates and cointegration between variables do not indicate the direction of the causal relationships between those variables. We thus use the panel causality test developed by Dumitrescu & Hurlin (2012) to investigate these correlations and offer crucial information to policymakers so they may implement sensible, efficient policies. This test combines Granger tests that are taken into account for each cross-sectional unit. The Dumitrescu-Hurlin causality test is flexible enough to be employed in both scenarios, N > T and N < T. (Haseeb et al., 2018). More crucially, the Dumitrescu-Hurlin method can address the problem of cross-sectional dependency. (Dogan & Seker, 2016a). The following linear model (equation (13)) is used to examine the causal relationship between variables X and Y:

                              (13)

where ∝_i is a constant term, γ_i^(k) Indicates the lag parameter, K denotes the lag length, and β_i^(k) is the slope coefficient. γ_i^(k) and β_i^(k) present the differences between cross-section units.

The null hypothesis and the alternative hypothesis of this test are presented as follows:

The alternative hypothesis postulates that the causal association exists in at least one subgroup, in contrast to the null hypothesis, which states that there is no causal relationship in the panel. According to Dumitrescu & Hurlin (2012), the Wald statistic (equation (14)), which is generated by averaging the values of the individual Wald statistics for each cross-sectional unit, may be used to examine these hypotheses:

                                                                                                                                (14)

In the case of T > N, the following statistical test (equation (15)) is suggested:

                                                                                                                         (15)

EMPIRICAL RESULTS AND DISCUSSION

The Pesaran and Yamagata slope homogeneity test was conducted on the panels. The null hypothesis states “homogeneous slope coefficients.” All panels exhibit delta estimations that are significant at the 1% level. The study employs heterogeneous panel techniques to tackle the issue of heterogeneous slopes, despite the diversity of the sample countries.

Table 5: Results of the Slope homogeneity tests.

H₀: slope coefficients are homogenous. ^a represents statistical significance at 1%.

∆ ̅ and ∆ ̅adj represent the “simple” and “mean-variance bias adjusted” slope homogeneity tests, respectively (Pesaran, Yamagata. 2008. Journal of Econometrics).

∆_HAC and 〖(∆〗_(HAC)adj )represent the “Heteroscedasticity and Autocorrelation Consistent” versions of “simple” and “mean-variance bias adjusted” slope homogeneity tests, respectively (Blomquist, Westerlund. 2013. Economic Letters).

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

Table 6 illustrates cross-sectional dependence among the factors examined in the study, leading to rejecting the null hypothesis of independence. This reinforces the notion of interdependence among nations across various stages of the demographic dividend and highlights global connections across multiple factors. The findings affirm the interdependence of countries in the Pre, Early, Late, and Post demographic dividend stages, as well as globally concerning LHDI, LPOP, LGDP, LENG, LFLFPR, and LMLFPR.

Table 6: Estimates of Cross-Sectional Dependence Tests

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

Table 7 presents the outcomes of second-generation panel unit root tests, suitable for data with heterogeneity and cross-sectional dependence problems (CADF and CIPS). The findings indicate that the variables LHDI, LPOP, LGDP, LENG, LFLFPR, and LMLFPR are stationary at the first difference but non-stationary at their level. This suggests that all the variables examined in the study are integrated at level 1 across every panel.

Table 7: Results of the CADF and CIPS panel unit root tests.

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

Table 8 displays the results of the Westerlund cointegration test for both linear and nonlinear models. The Gt and Pt statistics null hypothesis in these models is rejected at the 1% significance level in the Pre, Early Dividend, and Global panels. Both models exhibit long-term stability concerning the variables under study, indicating cointegration in at least one cross-sectional unit. However, only the nonlinear model demonstrates long-term stability or cointegration in the Late and Post dividend panels.

Table 8: Results of the Westerlund (2007) cointegration test.

Ho: No cointegration

 “a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

The results of the Driscoll-Kraay standard error regression are shown in Table 9 and are classified into linear and nonlinear models. The results of the linear model can be summarised as follows.

• While previous studies have suggested that population and population growth rates reduce HDI (Zgheib et al., 2006), (Arisman 2018), (Akbar et al., 2021) and (Tripathi, 2019)This study found a positive and significant relationship between a country’s total population and HDI across all demographic stages Also, a positive and meaningful relationship was found between a country’s per-capita GDP and HDI across all demographic stages, with varying elasticities. The findings align with previous studies by (Wang et al., 2021), (Barus et al., 2021), (Humaira & Nugraha, 2018), (Khan et al., 2019), and (Arisman 2018) That indicate the positive impact of GDP on HDI. The findings emphasize the importance of considering a country’s demographic stage when evaluating the relationship between GDP and HDI. Energy usage (LENG) significantly and positively impacts HDI across pre- and early-dividend panels, but it also hurts the global panel, according to Ouedraogo (2013). However, the effect of ENG on HDI is not significant at the Late and Post dividend stages. Wang et al.  (2021) suggest that renewable energy may increase HDI.
• The linear model estimates indicate that the elasticities between FLFPR and HDI vary at different stages of the demographic dividend. During the pre-demographic dividend stage, the elasticity is 0.129, implying that an increase in FLFPR by one results in a rise in HDI by 0.129. The elasticities are 0.77 and -0.104 at the late and post-demographic dividend stages, respectively. The elasticity in the global panel, which encompasses all countries, is 0.012. However, FLFPR is insignificant in the early demographic dividend panel, indicating it does not significantly impact HDI at that stage. Overall, the findings suggest that enhancing FLFPR can lead to an improvement in HDI. The elasticities of MLFPR on HDI are -0.57, -0.569, and -0.12 at the pre-, late-, and post-demographic dividend stages, respectively, and -0.112 globally. An improvement in MLFPR may result in a reduction in HDI. However, FLFPR and MLFPR are insignificant in the early demographic dividend panel, signifying that they do not significantly impact HDI at that stage.
• In various stages of demographic dividends, a linear model was employed to examine the impact of different factors on the HDI. In the Pre-demographic dividend stage, the model indicated that POP, GDP, ENG, and FLFPR enhance HDI, while MLFPR diminishes it. During the early demographic dividend stage, POP, GDP, and ENG contribute positively to HDI, whereas male and female participation rates have no significant effect. In the Late demographic stage, POP, GDP, and FLFPR continue to enhance HDI, but ENG and MLFPR have a negative impact. In the post-dividend stage, POP, GDP, and FLFPR again improve HDI, but MLFPR reduces it. The effect of ENG is deemed insignificant. POP, GDP, ENG, and FLFPR can elevate global HDI, while the impacts of MLFPR are minimal at 5%. The linear model accounts for 65%, 68%, 78%, 78%, and 83% of the variation in HDI at the pre-, early-, late-, and post-dividend and global panels, respectively.

The estimates of the nonlinear model of the Driscoll-Kraay standard errors regression are present in Table 11, and it highlights that:

The nonlinear model examines how female labor force participation dynamics affect the demographic dividend stages and their impact on HDI. FLFPR has a U-shape impact on HDI during the pre-and post-stage and an inverse U-shape during the Early and Late stages. On a global scale, FLFPR has an Inverse U-shape impact on HDI. The estimates reveal that at pre-, early-, late-, and post-demographic dividend stages, the coefficients of LFLFPR and LFLFPR2 are -0.986 and 0.370, 0.398 and -0.130, 1.409 and -0.404, and -1.308 and 0.405, respectively, with a 95% confidence.

The results of the Driscoll-Kraay standard errors regression global panel estimate in Table 11, along with the estimates of linear and nonlinear models, allow the following findings to be drawn:

• According to the linear model, the elasticities of POP, GDP, ENG, FLFPR, and MLFPR are 0.009, 0.091, 0.087, 0.012, and -0.112, respectively. LFLFPR exhibits an inverse U-shaped distribution on HDI globally—the effect of MLFPR on HDI is insignificant at the 5% level.

Table 9: Long Run Estimates

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

Table 9 also presents the outcomes of linear and nonlinear models employing Newey-West standard error regression to evaluate the reliability of Driscoll-Kraay standard error regression estimates. While the coefficients match those from Driscoll-Kraay, the t-statistics are notably higher. The explanatory variables prove significant according to both methods.

Population and GDP enhance HDI across all dividend panels and globally, with the Late dividend stage energy usage anticipated to improve HDI. The FLFPR positively impacts HDI in every study panel, displaying a U-shaped pattern regarding FLFPR’s effect on HDI at the pre-and post-dividend stages yet showing an inverse U-shape globally as well as during the Early and Late dividend stages. Conversely, MLFPR decreases HDI in all panels except during the Early dividend stage.

Table 10 illustrates the Dumitrescu-Hurlin panel non-causality test results, which indicate bidirectional causality among POP, GDP, ENG, and FLFPR about HDI across all study panels. Unlike the Post dividend panel, MLFPR also demonstrates bidirectional linkage with HDI. GDP, ENG, FLFPR, and MLFPR present bidirectional causality with POP in all study panels, with ENG, FLFPR, and MLFPR exhibiting significant bidirectional patterns amongst themselves. Overall, these findings suggest interdependency among the study variables across all panels.

Table 10: Results of the Dumitrescu Hurlin Panel Causality Test

“a “p<.01, “b “p<.05, “c “p<.1

Authors calculations

CONCLUSIONS AND POLICY RECOMMENDATIONS

Even though the overall population significantly impacts the HDI across all stages of demographic dividends, its elasticity is considered only during the pre-demographic dividend phase. This stage features many young individuals engaged in education or economic activities, which can enhance HDI. Contrary to previous studies suggesting that population and population growth rates may negatively impact HDI, the findings of this study indicate that a larger population can improve HDI at all stages of demographic dividends and globally. The analysis shows a positive correlation between per-capita GDP and HDI across all study panels, with varying elasticities based on a country’s demographic stage. These findings emphasise the importance of considering demographic phases when analysing the connection between population and HDI, with implications for decision-makers focused on sustainable development and human welfare. Except during the post-demographic dividend stage, the study also found that energy use has a significant and varied effect on HDI based on a country’s demographic stage. These results highlight the importance of considering a country’s demographic stage when examining the link between energy consumption and HDI.

A study identified a long-run relationship between FLFPR and HDI. Additionally, it was discovered that the impact of FLFPR on HDI varied depending on the country’s demographic stage. Although the study found that FLFPR significantly affects HDI globally, its influence was considerable during the late and post-dividend stages. In contrast, it had an insignificant impact during the early dividend stages. FLFPR most favourably impacted HDI in the pre-demographic dividend stage. The relationship between FLFPR and HDI in different demographic dividend stages has been better understood using a nonlinear model. The findings suggest that the impact of FLFPR on HDI is not linear and varies with the stage of demographic change. The study reveals a U-shaped relationship between FLFPR and HDI in the pre and post-dividend stages, indicating that moderate FLFPR levels can enhance HDI. In contrast, both high and low FLFPR levels may adversely affect it. However, in the early and late dividend stages, FLFPR has an inverse U-shaped influence on HDI, implying that excessively low and excessively high FLFPR levels can harm HDI. Overall, FLFPR has an inverse U-shaped effect on HDI, and the cointegration study confirms the long-term stability of the nonlinear model across all study panels. These outcomes underscore the significance of considering the nonlinear relationship between FLFPR and HDI alongside the demographic stage when devising strategies to enhance HDI.

During the pre and late-dividend stages, the relationship between MLFPR and HDI shows high elasticities, indicating that enhancing MLFPR may significantly reduce HDI. These results hold important implications for policymakers aiming to promote human development. Findings suggest that increasing male labour force participation could negatively affect people’s welfare, particularly during the early and late stages of the demographic dividend cycle.

According to the Dumitrescu-Hurlin panel non-causation test, there is a two-way causal relationship between the independent variables (POP, GDP, ENG, FLFPR, and MLFPR) and HDI in all study panels, except for the post-dividend panel, where MLFPR does not show a significant causal connection with HDI. Additionally, all study panels indicate a mutual interconnection between GDP, ENG, FLFPR, and MLFPR with POP, suggesting that these variables are interdependent. ENG, FLFPR, and MLFPR exhibit a significant two-way causal association in all study panels. These findings imply that the research variables significantly impact each other and are highly interrelated.

Figure 2: Graphical summary of the findings

Pre-demographic dividends stage	Early demographic dividends stage	Late demographic dividends stage	Post demographic dividends stage	Global

Bidirectional causality

Created by the Author

Figure 1 graphically summarizes findings regarding the global, pre-, early-, late-, and post-demographic dividend panel estimations. In the pre-, post-, and global panels, FLFPR, POP, ENG, and GDP indicate long-term HDI improvements. However, MLFPR reduces HDI in the long run. All independent variables show bidirectional causality with HDI. According to the estimates of the nonlinear model, the relationship between FLFPR and HDI is U-shaped in the pre- and post-dividend panels. FLFPR and HDI exhibit an inverse U-shaped distribution in the early, late, and global panels. These findings help policymakers create more efficient policies to improve HDI by optimizing a county’s FLFPR. FLFPR, POP, and GDP improve HDI at all DD stages and globally. MLFPR improves HDI only at the Early DD stage. Based on these new insights, policymakers can create more valid policies to improve HDI by optimizing FLFPR.

The study found that a country’s HDI is influenced by its population, GDP, energy use, and gender-specific labour force participation rates, with this impact varying based on the country’s demographic stage. Policymakers can leverage this fresh evidence to improve a country’s HDI according to its demographic dividend stage. FLFPR significantly enhances HDI at all demographic dividend stages globally. Thus, policies should be developed to increase female employment and participation rates. Furthermore, bidirectional causality confirms that improvement in HDI boosts a country’s FLFPR. Optimising FLFPR advances human development by empowering women and contributes to achieving the SDGs. Finally, the study concludes that FLFPR is positively associated with sustainable development through enhancing human development.

REFERENCES

1. Ahmed, S. A., Cruz, M., Quillin, B., & Schellekens, P. (2016). Demographic change and development: Looking at challenges and opportunities through a new typology. World Bank: Policy Research Working Paper, WPS7893(November), 1–40. https://doi.org/10.1596/1813-9450-7893
2. Akbar, M., Hussain, A., Akbar, A., & Ullah, I. (2021). The dynamic association between healthcare spending, CO2 emissions, and human development index in OECD countries: evidence from panel VAR model. Environment, Development and Sustainability, 23(7), 10470–10489. https://doi.org/10.1007/s10668-020-01066-5
3. Appiah, E. N. (2018). Female Labor Force Participation and Economic Growth in Developing Countries. Global Journal of Human-Social Science: E Economics, 18(2), 175–192. http://wol.iza.org/articles/female-labor-force-participation-in-developing-countries
4. (2018). Determinant of Human Development Index in ASEAN Countries. JSignifikan: Jurnal Ilmu Ekonomi., 7(1), 113–122. https://doi.org/doi: http//dx.doi.org/10.15408/sjie.v7i1.6756
5. Bailey, N., Kapetanios, G., & Pesaran, and M. H. (2019). Exponent of Cross-sectional Dependence for Residuals. The Indian Journal of Statistics, 81, 46–102. https://doi.org/10.1007/s13571-019-00196-9
6. Balakrishnan, R., & Dharmaraj, K. (2018). Women and sustainable development goals. In The Oxford Handbook of Women, Peace, and Security. https://doi.org/10.1093/oxfordhb/9780190638276.013.53
7. Baloch, M. A., Zhang, J., Iqbal, K., & Iqbal, Z. (2019). The effect of financial development on ecological footprint in BRI countries: evidence from panel data estimation. Environmental Science and Pollution Research, 26(6), 6199–6208. https://doi.org/10.1007/s11356-018-3992-9
8. Barus, E. N., Tarmizi, H., & . R. (2021). Analysis of Factors Affecting Human Development Index in Special Regional of Yogyakarta. International Journal of Research and Review, 8(4), 161–170. https://doi.org/10.52403/ijrr.20210422
9. Bekhet, H. A., & Othman, N. S. (2017). Impact of urbanization growth on Malaysia CO2 emissions: Evidence from the dynamic relationship. Journal of Cleaner Production, 154, 374–388. https://doi.org/10.1016/j.jclepro.2017.03.174
10. Blomquist, J., & Westerlund, J. (2013). Testing slope homogeneity in large panels with serial correlation. Economics Letters, 121(3), 374–378. https://doi.org/10.1016/j.econlet.2013.09.012
11. Choudhry, M. T., & Elhorst, P. (2018). Female labour force participation and economic development. International Journal of Manpower, 39(7), 896–912. https://doi.org/10.1108/IJM-03-2017-0045
12. Chudik, A., & Pesaran, M. H. (2015). Common correlated effects estimation of heterogeneous dynamic panel data models with weakly exogenous regressors. Journal of Econometrics, 188(2), 393–420. https://doi.org/10.1016/j.jeconom.2015.03.007
13. Dasic, B., Devic, Z., Denic, N., Zlatkovic, D., Ilic, I. D., Cao, Y., Jermsittiparsert, K., & Le, H. Van. (2020). Human development index in a context of human development: Review on the western Balkans countries. Brain and Behavior, 10(9). https://doi.org/10.1002/brb3.1755
14. Denney, J. M. (2015). Gender and the Sustainable Development Goals: Moving Beyond Women as a “Quick Fix” for Development. Governance and Sustainability Issue Brief Series:Brief 11, 7.
15. Dogan, E., & Seker, F. (2016a). Determinants of CO2 emissions in the European Union: The role of renewable and non-renewable energy. Renewable Energy, 94(2016), 429–439. https://doi.org/10.1016/j.renene.2016.03.078
16. Dogan, E., & Seker, F. (2016b). The influence of real output, renewable and non-renewable energy, trade and financial development on carbon emissions in the top renewable energy countries. Renewable and Sustainable Energy Reviews, 60, 1074–1085. https://doi.org/10.1016/j.rser.2016.02.006
17. Driscoll, J. C., & Kraay, A. C. (1998). Consistent covariance matrix estimation with spatially dependent panel data. Review of Economics and Statistics, 80(4), 549–559. https://doi.org/10.1162/003465398557825
18. Dumitrescu, E. I., & Hurlin, C. (2012). Testing for Granger non-causality in heterogeneous panels. Economic Modelling, 29(4), 1450–1460. https://doi.org/10.1016/j.econmod.2012.02.014
19. Foster, E. A. (2016). Gender and sustainable development. Handbook on Gender in World Politics, 446–454. https://doi.org/10.4337/9781783470624.00064
20. Gelard, P., & Abdi, A. (2015). Evaluating the effect of gender inequality on economic growth in countries with high human development index. European Online Journal of Natural and Social Sciences Special Issue on New Dimensions in Economics, 4(1), 1714–1724. http://www.european-science.com
21. Haque, M. I., & Khan, M. R. (2019). Role of oil production and government expenditure in improving human development index: Evidence from Saudi Arabia. International Journal of Energy Economics and Policy, 9(2), 251–256. https://doi.org/10.32479/ijeep.7404
22. Haseeb, A., Xia, E., Danish, Baloch, M. A., & Abbas, K. (2018). Financial development, globalization, and CO2 emission in the presence of EKC: evidence from BRICS countries. Environmental Science and Pollution Research, 25(31), 31283–31296. https://doi.org/10.1007/s11356-018-3034-7
23. Hashem Pesaran, M., & Yamagata, T. (2008). Testing slope homogeneity in large panels. Journal of Econometrics, 142(1), 50–93. https://doi.org/10.1016/j.jeconom.2007.05.010
24. Hashemizadeh, A., Bui, Q., & Kongbuamai, N. (2021). Unpacking the role of public debt in renewable energy consumption: New insights from the emerging countries. Energy, 224, 120187. https://doi.org/10.1016/j.energy.2021.120187
25. Hoechle, D. (2007). Robust standard errors for panel regressions with cross-sectional dependence. Stata Journal, 7(3), 281–312. https://doi.org/10.1177/1536867×0700700301
26. Humaira, U. H., & Nugraha, J. (2018). Analysis of Factors Affecting the Human Development Index in West Kalimantan Province using Data Panel Data Regression. Journal of Sciences and Data Analysis, 18, 97–105. https://doi.org/10.20885/eksakta.vol18.iss2.art2
27. Ivanova, F. J. A. and G. S. (1999). An Assessment of the Measurement Properties of the Human Development Index. Social Indicators Research, 46(2), 157–179.
28. Khan, N. H., Ju, Y., & Hassan, S. T. (2019). Investigating the determinants of human development index in Pakistan: an empirical analysis. Environmental Science and Pollution Research, 26(19), 19294–19304. https://doi.org/10.1007/s11356-019-05271-2
29. Kongbuamai, N., Bui, Q., Yousaf, H. M. A. U., & Liu, Y. (2020). The impact of tourism and natural resources on the ecological footprint: a case study of ASEAN countries. Environmental Science and Pollution Research, 27(16), 19251–19264. https://doi.org/10.1007/s11356-020-08582-x
30. Lee, R. (2003). The demographic transition: Three centuries of fundamental change. Journal of Economic Perspectives, 17(4), 167–190. https://doi.org/10.1257/089533003772034943
31. Lee, R. D., & Mason, A. (2006). What is the demographic dividend? Finance and Development, 43(3), 16–17.
32. Lee, S.-H., & Mason, A. (2019). Demographic dividends. Policies based on understanding population dynamics will help countries achieve the SDGs. Sustainable Development Goals 2019, 42–43. https://www.sustainablegoals.org.uk/wp-content/uploads/2019/06/042-043-SDGs-LEE.pdf
33. Madhusudan Ghosh. (2018). Gender Equality, Growth and Human Development in India. Asian Development Perspectives, 1(June), 68–87. https://doi.org/10.22681/ADP.2018.9.1.68
34. John Wilmoth, M. C. M. and M. L. B. (2022). Why population growth matters for sustainable development. POLICY BRIEF NO 130, Future of the World, United Nations Department of Economic and Social Affairs, 130, 1–4. https://www.un.org/development/desa/ pd/sites/www.un.org. development.desa.pd/files/undesa_pd_2022_policy_brief_population_growth.pdf
35. Bailey, G. Kapetanios, M. H. P. (2015). EXPONENT OF CROSS-SECTIONAL DEPENDENCE: ESTIMATION AND INFERENCE. Journal of Applied Econometrics, 21(August 2012), 1–21. https://doi.org/10.1002/jae
36. Newey, W. K., & West, K. D. (2010). West, 1987, A simple positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix estimator. Econometrica, 55(3), 703–708. http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:A+Simple+,+Positive+Semi-Definite+,+Heteroskedasticity+and+Autocorrelation+Consistent+Covariance+Matrix#2
37. Ouedraogo, N. S. (2013). Energy consumption and human development: Evidence from a panel cointegration and error correction model. Energy, 63, 28–41. https://doi.org/10.1016/j.energy.2013.09.067
38. A. V. B. Swamy. (1970). Efficient Inference in a Random Coefficient Regression. Econometrica, 38(2), 311–323.
39. Pesaran, M. H. (2004). General Diagnostic Tests for Cross Section Dependence in Panels. Iza, 1240, (August), 1–42. http://www.dspace.cam.ac.uk/handle/1810/446
40. Pesaran, M. H. (2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics, 22(2), 265–312. https://doi.org/10.1002/jae.951
41. Peter C. Phillips and Donggyu Sul. (2003). Dynamic panel estimation and homogeneity testing under cross-section dependence ∗. Econometrics Journal (2003), 6(June 2002), 217–259.
42. Rahman, M. M., & Alam, K. (2022). The roles of globalization, renewable energy and technological innovation in improving air quality: Evidence from the world’s 60 most open countries. Energy Reports, 8, 9889–9898. https://doi.org/10.1016/j.egyr.2022.07.165
43. Sagar, A. D., & Najam, A. (1998). The human development index: A critical review. Ecological Economics, 25(3), 249–264. https://doi.org/10.1016/S0921-8009(97)00168-7
44. Shah, S. (2016). Determinants of Human Development Index: A Cross-Country Empirical Analysis. International Journal of Economics and Management Studies, 3(5), 43–46. https://doi.org/10.14445/23939125/ijems-v3i5p106
45. Shahbaz, M. (2013). Does financial instability increase environmental degradation? Fresh evidence from Pakistan. Economic Modelling, 33, 537–544. https://doi.org/10.1016/j.econmod.2013.04.035
46. Sinha, A., & Sen, S. (2016). Atmospheric consequences of trade and human development: A case of BRIC countries. Atmospheric Pollution Research, 7(6), 980–989. https://doi.org/10.1016/j.apr.2016.06.003
47. Taheri, E., Güven Lisaniler, F., & Payaslioğlu, C. (2021). Female labour force participation: What prevents sustainable development goals from being realised in Iran? Sustainability (Switzerland), 13(21). https://doi.org/10.3390/su132111918
48. Tran, N. Van, Tran, Q. Van, Do, L. T. T., Dinh, L. H., & Do, H. T. T. (2019). The trade-off between environment, energy consumption and human development: Do levels of economic development matter? Energy, 173, 483–493. https://doi.org/10.1016/j.energy.2019.02.042
49. Tripathi, S. (2019). Urbanization and Human Development Index : Cross-country evidence. Munich Personal RePEc Archive, 97474.
50. (2016). Human Development Report: Human Development for Everyone. In United Nations Development Programme. http://hdr.undp.org/sites/default/files/2016_human_development_report.pdf
51. Ustabaş, A., & Gülsoy, T. Y. (2017). The Relationships Between the Female Labor Force Participation Rate and Economic Development: A Correlation Analysis for Turkey. International Conference on Eurasian Economies 2017, 2016, 104–113. https://doi.org/10.36880/c08.01912
52. Wang, Z., Bui, Q., Zhang, B., Nawarathna, C. L. K., & Mombeuil, C. (2021). The nexus between renewable energy consumption and human development in BRICS countries: The moderating role of public debt. Renewable Energy, 165, 381–390. https://doi.org/10.1016/j.renene.2020.10.144
53. Wang, Z., Danish, Zhang, B., & Wang, B. (2018). Renewable energy consumption, economic growth and human development index in Pakistan: Evidence from simultaneous equation model. In Journal of Cleaner Production (Vol. 184). Elsevier Ltd. https://doi.org/10.1016/j.jclepro.2018.02.260
54. Westerlund, J. (2007). Testing for error correction in panel data. Oxford Bulletin of Economics and Statistics, 69(6), 709–748. https://doi.org/10.1111/j.1468-0084.2007.00477.x
55. Zgheib, P. W., Ahmed, Z. U., Beldona, S., & Gebara, V. (2006). The impact of population growth on Human Development Index: a comparative analysis of Middle Eastern countries. World Review of Science, Technology and Sustainable Development, 3(3), 258–269. https://doi.org/10.1504/WRSTSD.2006.010226 

APPENDIX

List of countries selected for the study panels.

Table A1: List of 20 selected countries for the pre-demographic dividend panel.

Source: Created by the author based on the WDI (2022), https://data.worldbank.org/country/V1.

Table A2: List of 41 selected countries for the early-demographic dividend panel.

Source: Created by the author based on the WDI (2022), https://data.worldbank.org/country/early-demographic-dividend

Table A3: List of 29 selected countries for the late-demographic dividend panel.

Source: Created by the author based on the WDI (2022), https://data.worldbank.org/country/late-demographic-dividend

Table A4: List of 28 selected countries for the post-demographic dividend panel.

Source: Created by the author based on the WDI (2022), https://data.worldbank.org/country/post-demographic-dividend