Introduction

The female labour force participation (FLFP) rate is crucial to a nation’s socioeconomic development. In general, a high FLFP signifies an improvement in a nation’s social and economic standing, leading to the empowerment of women, which in turn promotes equity and boosts the utilization of human potential, hence fostering economic growth and reducing poverty (Andlib and Khan, 2018). Women’s economic participation is also connected to favourable outcomes, such as improved nutrition, educational attainment, and greater participation in household decision-making. Nonetheless, we must recognize the quality of the employment and the compensation it provides. Various contextual factors, such as mobility, segregation, and gender norms about reproduction and reproductive labour, affect women’s entry into the workforce (Andlib et al., 2022).

Digital technologies are causing a rapid transformation in the world of work. Today, digitalization has permeated almost all major economic sectors (Silva, 2022). From software developers and programmers to domestic workers on digital labour platforms to market vendors and microentrepreneurs who use digital tools to reach customers, digital labour encompasses a wide range of occupations. The COVID-19 pandemic has accelerated digitalization as work and livelihoods have shifted online. However, it also exposed and exacerbated inequalities between the global North and South along gender, race, caste, and class lines. Common reasons for this digital divide include limited access to digital infrastructure, low digital literacy, and repressive sociocultural norms (Prates and Barbosa, 2020; Sarfraz et al., 2021).

At the commencement of the COVID-19 pandemic, policy experts and scholars anticipated that the pandemic and succeeding restriction strategies would expedite digitalization with a potentially significant indication for labour markets and efficiency. Most of the workers had to shift from working in the office to working from home, and contact-concentrated economic activities were limited (Woźniak-Jęchorek and Marchewka-Bartkowiak, 2023). Consequently, numerous firms had to adapt to remote work and broaden their activities online, which necessitated variation in functioning, management and maybe a speedy investment in information and communication technology (ICT). Throughout the COVID-19 pandemic, these forms of digitalization that permit businesses to function remotely may have encouraged employment and labour efficiency; there were expectations that they would uplift firms’ and workers’ efficiency in the longer term (Asongu and Odhiambo, 2022). Contrary to it, hopes concerning the long-term influence on the labour market were more mixed with worry that digitalization could disturb the incomes of low and middle-skilled workers.

Governments in BRICS economies have embraced digitalization in the hope that it will boost productivity and competitiveness while also creating jobs. Given the high and rising unemployment, the latter is crucial. Indeed, international development agencies have widely promoted digitalization as a policy path towards sustainable, inclusive, and equitable economic growth, potentially improving women’s social and economic outcomes (UN Women, 2020). Huws (2014), on the other hand, contends that digital innovation has facilitated the concentration of capital across industries and geographies, increasing their monopoly power in a context where states’ regulatory capacity is already precarious. Workers’ bargaining power has been severely eroded as surplus value is increasingly derived from value extraction rather than commodity production (Sarfraz et al., 2022).

The digital economy in BRICS economies is rapidly expanding, helping to boost growth and expand economic opportunities at a time when global productivity growth has been disappointing. Digital platforms are assisting micro-businesses in low and middle-income countries in accessing new, and in some cases international, markets and supplementing traditional sources of income (Rani et al., 2021). The rapid increase in mobile phone availability and use in the region has assisted in lowering the costs of obtaining information and other transaction costs, lowering the costs of money transfer and financial services, improving access to credit, and helping women in balancing their work and family lives (Herman, 2020). Automation and skill-biased technological change have increased the demand for ‘brains’ relative to ‘brawn’ in both developed and developing economies, allowing women to close participation and pay gaps (Cirillo et al., 2021). While the Internet has increased women’s workforce participation in developed economies by enabling teleworking and flexible work arrangements, it has also reduced the time spent on household labour (Akarsu, 2023).

The relationship between female labour force participation and the total fertility rate has received much attention in the demographic and economic literature. It has also been debated in various regions and stages of development worldwide. In this regard, the presence of children impacts the mother’s activity, primarily if the mother works. Mothers with small children are traditionally thought to have low labour force attachment (Bhalotra et al., 2021). Female labour force participation will also have an impact on family size decisions. As a result, the discussion of female labour force participation cannot be separated from the discussion of fertility, as the two are inextricably linked. Generally, female labour force participation is high in countries with low fertility rates (Buyukkececi and Engelhardt, 2021).

The ongoing research confabulates the influence of digitalization on female labour force participation in BRICS nations. We have chosen BRICS nations for so many valid reasons. First, these nations are emerging nations and focusing a lot on digitalization. China and India’s services sectors are making their mark in the world. Secondly, these chosen nations occupy almost 42 per cent of the world’s population, and these nations have a 24 per cent share of the world’s GDP. Further, on average, these nations occupy a 16 per cent share of the world’s trade. The FLFP rate is 46 per cent in the chosen block, and China has the highest (World Development Indicators). In contrast, India has the lowest FLFP Russia has the highest number of people in the population who are using the Internet. India has the lowest percentage of the population using the Internet in this block.

Table 1 Cross-sectional dependence analysis.
Full size table

The present analysis assesses the impact of digitalization, education, fertility and GDP on FLFP in BRICS economies. The study adds to prior work in many ways. It is the first-ever study on the digitalization and FLFP nexus for BRICS. The study uses new and comprehensive econometrics techniques for rigorous results. The study has included five emerging (B—Brazil, R—Russia, I—India, C—China, S—South Africa) economies with economic and geographic importance in the respective regions and a rich natural resource base. The study provides a few important policy implications which can be generalized for any other developing country or group of countries. Therefore, specific research questions which are assessed by our study are as follows: (1) Does digitalization help to enhance the female labour force participation rates in BRICS nations? (2) Does education increase the FLFP? (3) Does an increase in GDP facilitate FLFP in the chosen nations?

The rest of the study is organized as follows: “Literature review”, “Data sources and methodology”, ”Results and discussions” and “Conclusions and policy insights”.

Literature review

This section will briefly give an overview of the selected variables, digitalization, education, fertility and growth with FLFP. The empirical evidence is taken from different economies around the globe.

Suhaida et al. (2013) illuminate that ICT enacts as a driving force in escalating a woman’s decision to work. ICT permits workers to do their job more productively, owing to greater flexibility, allowing them to work from home in Malaysia. Nikulin (2017) ascertains that when the analysis takes the evidence from 60 developing countries, there is an existence of a direct association between ICT and FLFP. Efobi et al. (2018) evaluate that ICT is considered a dominant factor in boosting FLFP with increasing magnitude: mobile phone penetration, internet penetration, and fixed broadband subscriptions in the case of 48 African countries. Asongu and Odhiambo (2018) depict that ICT functions as a productive strategy in altering FLFP in the context of 48 African countries. ICT adjusts the financial approach to instigate approving impact on female economic contribution. Samargandi et al., (2019) determine that ICT impedes FLFP in the context of Saudi Arabia. It is also revealed that financial development modifies the unfavourable effect of ICT dispersion. Jain (2021) culminates that ICT tends to have a favourable influence on FLFP in the case of India. Moreover, the positive influence of ICTs on female employment is bound to the portion of females in the proficient workforce. Hafeez et al. (2020) evaluate in the case of selected South Asian countries ICT support in improving FLFP. As (ICT) is reviewed as one of the fundamental drivers concerning women’s empowerment. Ngoa and Song (2021) demonstrate that in the case of 48 African countries, ICT plays a critical role in encouraging FLFP The impact of ICT on female employment in Africa is predominant in the industrial sector. Tüzemen et al. (2021) interpret that in the case of Turkey, there is no interrelation between ICT and FLFP. The analysis recommends that developing countries require to prosper in other aspects to gain from the opportunities that ICT provides to empower society. Viollaz and Winkler (2022) elucidate that in the case of Jordan, ICT assists in expanding FLFP because internet adoption expands online job search. Women with higher levels of education undergo a rise in employment, attaining an internet approach. Galperin and Arcidiacono (2021) exemplify that the interconnection between employment and internet use is greater among women than men in four Latin American countries. Asongu and Odhiambo (2022) exhibit that in the context of sub-Saharan African nations, ICT in broadband subscriptions emboldens the female employment rate.

Faridi et al. (2009) illuminate that education facilitates enhanced Female Labour Force participation. The coefficients of all the levels of education except primary education up to the middle level are significant in the Logit Regression equation. It is revealed that educated females are leading toward economic growth. Ince (2010) illustrates that when the analysis takes evidence from Turkey, education is a critical factor in influencing female employment as education directly influences female employment. Khanie (2019) dictates that in the case of Botswana, education functions as a substantial element in improving FLFP as women with higher education are more likely to be wage employed. In contrast, those with lower to no education are less likely to be wage employed. Bhalla and Meher (2019) interpret that even though women’s educational level is constantly increasing, the employment rates have not improved at a similar speed in Kerala. Women’s individual job preferences, family constraints, lack of adequate skills, and discouraged worker’s effects are some causes of high unemployment among females. Onyeke and Ukwueze (2022) deduce in the context of Nigeria the primary school education enrolment for females secondary school education enrolment for females and have a long-run positive impact on the female labour force participation rate. So the government should arrange all social needs to motivate schooling among Nigerian females and all citizens.

Ukil (2015) regulates that fertility has an unfavourable influence on FLFP as fertility is endogenous to female labour participation in the case of Britain. Shittu and Abdullah (2019) clarify that fertility inhibits FLFP as it has a negative impact on FLFP in the case of ASEAN-7 countries. Nazah et al. (2021) consider that in the case of 39 Asian countries, fertility was negatively significant on female labour participation in the short run but not in the long run due to its close relationship with female labour force participation. Bawazir et al. (2022) enlighten that fertility obstructs the females’ labour force participation rate in the case of Middle East countries. Bloom et al. (2009) discovered a significant negative effect of fertility on female labour force participation in 97 countries between 1960 and 2000, using abortion as an instrument and simulation variable. Nakagaki (2018) examined 176 countries using a fixed effect and a random effect and found a correlation that shifted from negative to positive by the 1990s in OECD countries. However, no clear pattern was found between female labour force participation and fertility rates in the Asian-Pacific region.

Mishra et al. (2010) enlighten that for G-7 nations total fertility rate persuades the economy towards curtailing female labour force participation as they are adversely associated. Baah-Boatenga et al. (2013) exhibit that education and fertility are effective mechanisms for escalating female labour force participation in Ghana. Abu Bakar et al. (2014) demonstrate that for 6 Asian nations, there is a presence of unfavourable connotation between the total fertility rate and female labour force participation rate. Hartani et al. (2015) elucidate that for 6 ASEAN nations total fertility rate alleviates female labour force participation. The female labour force participation rate’s highest adverse impact is perceived for Indonesia and smallest for Thailand. Tanaka et al. (2020) deduce that in Bangladesh, education tends to have a favourable influence on female labour force participation. Moreover, education leads to a favourable influence on sanitation control and children’s health.

Chapman (2015) explains that in the case of 20 countries in MENA, there is the appearance of a U-shaped relationship between economic growth and female labour force participation rates. Their transformation regarding the bottom of the U-shaped curve demonstrates the low female labour force contribution rates. Lechman and Kaur (2015) inspect that in the case of 162 countries, a U-shaped link exists between economic growth and FLFP. However, only in the case of low-income countries the U-shaped feminization hypothesis was not positively certified. Belke and Bolat (2016) resolve that when the analysis takes the evidence from 148 developed and developing countries, the outcome validates the U-shape association between economic development and female labour participation. Tasseven (2017) expounds that gross domestic product plays a pivotal role in stimulating female labour force participation positively in the context of G8 countries because male and female labour force participation is required to achieve economic development. Sasongko et al. (2020) discover that in the context of 34 provinces in Indonesia female labour force participation rate is not altered by economic growth. The Indonesian government is obliged to examine minimum wages, education, work age, and work experience as policy tools to raise the female labour force participation rate. Beton Kalmaz (2023) ascertained that education facilitates increasing female labour force participation in Turkey as they are favourably connected.

We have done much prior work but have not found a comprehensive study on the specific and chosen connotations for BRICS economies. Therefore, this study is the first-ever attempt in this regard. By looking at the prior work on the interconnections between digitalization, education, fertility and FLFP, we can infer that there needs to be more literature on the said linkages for the group of five most significant economies in the region, i.e., BRICS. Thus, the present study attempts to illustrate the association between digitalization, education and FLFP. The study also incorporated other critical economic indicators. The empirical outcomes of the present research will pave the way for policymakers to formulate suitable policies for the developing economies included in the sample.

Methods

Theoretical framework and data

Even though we could not find any comprehensive study for the chosen nations based on a theoretical underpinning we have established from various notable works on the subject of interest (Hafeez et al., 2020; Onyeke and Ukwueze, 2022), we have developed our theoretical model, which we will discuss in this section. In the case of five selected BRICS economies, the current analysis evaluates the impact of digitalization and education on FLFP. Furthermore, it will also confabulate the impact of fertility and economic growth on FLFP. The period for the said analysis is from 1990 to 2020. We will have the following functional form to test the resource curse hypothesis.

$${\rm {FLFP}}_{it} = f\left( {{\rm {Digi}}_{it},{\rm {Edu}}_{it}} \right)$$
(1)

FLFP signify female labour force participation, Edu is years of education, Digi is digitalization. Our analysis incorporates two other essential economic indicators, i.e., GDP and fertility, to elucidate their influence on FLFP. Equation (1) “i” illustrates the cross-sections of Brazil, Russia, India, China and South Africa, whereas “t” is the time frame.

$${\rm {FLFP}}_{it} = f\left( {{\rm {Digi}}_{it},{\rm {Edu}}_{it},{\rm {Fert}}_{it},{\rm {GDP}}_{it}} \right)$$
(2)

Equation (3) reveals the regression form for our analysis. \(\phi _{it}\) is a cross-section specification term, and it is the error term.

$$\begin{array}{ll}{\rm {FLFP}}_{it} = \beta _{1it} + \beta _{2it}{\rm {Digi}}_{it} + \beta _{3it}{\rm {Edu}}_{it}\\ \qquad\qquad\,+\, \beta _{4it}{\rm {Fert}}_{it} + \beta _{5it}{\rm {GDP}}_{it} + \phi _{it} + v_{it}\end{array}$$
(3)

To reduce the skewness in the data, we have taken the natural logarithm of the selected variables. In Eq. (3), FLFP is the female labour force participation (World Development Indicators). GDP is the gross domestic product at constant dollars (2010) (WDI). Education stands for secondary school enrolment for women (WDI). Fertility is represented by the fertility rate. For digitalization, we have taken the number of internet users in the total population. Even though this number is gross and considers females and females, if the number of internet users increases, it will be evenly distributed in a household and the population.

We looked closely at the prior literature to specify the expected interconnections among selected variables.

$$\beta _{2it} = \frac{{\partial {\rm {FLFP}}}}{{\partial {\rm {Digi}}}}\, >\, 0$$

We have also included education in our empirical study. Looking at the prior literature, we expect that Edu positively influences FLFP.

$$\beta _{3it} = \frac{{\partial {\rm {FLFP}}}}{{\partial {\rm {Edu}}}}\, >\, 0$$

The interconnection between fertility and FLFP is expected to be positive.

$$\beta _{4it} = \frac{{\partial {\rm {FLFP}}}}{{\partial {\rm {Fert}}}}\, < \,0$$

In the end, we included GDP in our analysis; we expect a positive association between these two variables.

$$\beta _{5it} = \frac{{\partial {\rm {FLFP}}}}{{\partial {\rm {GDP}}}}\, > \,0$$

Methodology

Unit root tests

Nowadays, economies are interconnected on different grounds, for instance, economic, financial, and cultural. Consequently, oil price shocks, financial crises, pandemics, the interdependence of residuals, and unobserved common factors are connected with cross-section dependence (Csd). Our empirical outcomes are spurious and biased if we do not consider the issue of Csd. Thus, it is mandatory to consider the issue of Csd in our analysis. Here to cope with the issue of Csd, we have utilized Pesaran (2015) Csd test. It is always better to Csd test before testing the stationarity of the data. Based on the Csd statistics, we specify the unit root we will apply to our selected variables.

In panel data studies, we divided unit root tests into various generations based on different issues we have to come across in panel data estimation. For example, first-generation tests (Maddala and Wu, 1999; Choi, 2001; Levin et al., 2002) are more likely to be dealt with non-stationarity with homogeneous panels. The issues related to non-stationarity with heterogeneous panels are solved by Im et al. (2003). Furthermore, to address the issue of structural breaks, we prefer to apply a test given by Lluís Carrion‐i‐Silvestre et al. (2005); however, this test cannot tackle the issue of Csd.

In a nutshell, second-generation tests can deal with the issues of heterogeneity and Csd between the units, but these tests cannot consider the issue of structural breaks. To overcome the issue of structural breaks in panel data, we prefer to apply third-generation unit root tests because these tests consider the three most compelling issues of panel data, heterogeneity, Csd, and structural breaks. Thus, the present analysis will apply two tests: Bai and Carrion-I-Silvestre (2009) and Pesaran (2007).

Cointegration testing

The most compelling drawback with first-generation cointegration approaches (McCoskey and Kao, 1998; Larsson, 2001; Pedroni, 2004 and Westerlund, 2005) is that in the presence of Csd, these approaches are unable to offer unbiased estimates—size properties distortions. In addition, a few other approaches, including Kao et al. (1999) and Pedroni (2001), cannot overcome the issue of Csd. We aim to use the most appropriate and comprehensive cointegration test, which can give us rigorous estimates even in the presence of heterogeneity, Csd, and structural breaks. Therefore we consider heterogeneous estimation methods, Westerlund and Edgerton (2008) and Banerjee and Carrion‐i‐Silvestre (2017). Also, these methods have the edge over a few other methods, like, Westerlund (2007), which deals with the issue of Csd and heterogeneous slopes parameters. Still, it is unable to solve the problem of structural breaks in the panel data. Thus, this method may reject the null hypothesis of no cointegration even if there is a cointegration. Westerlund and Edgerton’s (2008) comprehensive method also elucidates the issues of Csd, autocorrelation, heterogeneous slopes, and structural breaks. However, in the present analysis, we have utilized Banerjee and Carrion‐i‐Silvestre’s (2017) test, which considers common correlated effects mean group. This approach has many merits in dealing with non-stationary data, heterogeneity, and weak and strong Csd.

Cross-sectionally augmented autoregressive distributed lags (CSARDL) model

As previously mentioned that panel data has to suffer from different issues. Moreover, countries have to confront various kinds of shocks, for example, economic disturbances and financial shocks, pandemics and natural disasters. These shocks cause the issue of Csd. If we do not incorporate these shocks into our analysis, we get biased results. To deal with the issue of slope heterogeneity and also Csd, we apply the most compelling approach CSARDL. We start with,

$$V_{i,t} = \mathop {\sum}\limits_{K = 0}^{n_w} {\vartheta _{K,i}V_{i,t - 1}} + \mathop {\sum}\limits_{K = 0}^{n_v} {\delta _{K,i}X_{i,t - 1} + \mu _{i.t}}$$
(4)

Equation (3) elucidates the autoregressive distributed lags model, but in the presence of cross-section dependence, it provides biased results. Nonetheless, in Eq. (4), we have used a cross-section average of each regressor. To overcome the issue of Csd, Eq. (5) gives the advantage to solve the unfitting inference related to the presence of threshold effects (Chudik and Pesaran, 2015).

$$V_{i,t}=\mathop{\sum}\limits_{K=0}^{n_{w}}\vartheta_{K,i}V_{i,t-1}+\mathop{\sum}\limits_{k=0}^{n_{v}}\delta_{K,i}X_{i,t-1}+\mathop{\sum}\limits_{k=0}^{n_{y}}{\acute{\gamma}}_{i},K{\bar{Y}}_{i,t-1}+\mu_{i.t}$$
(5)

In Eq. (5) \({\bar{Y}}_{i,t-1}={\bar{V}}_{i,t-1},{\bar{X}}_{i,t-1}\) specify the averages of the dependent variable, i.e. FLFP and independent variables Digi, Edu, Fert and GDP. NW, \(n_z\) and \(n_y\) are lags of each variable. Also, \(V_{i,t}\) represents the dependent variable and \(X_{i,t}\) represents the set of independent variables. Whereas \({\bar{Y}}\) refers to the cross-section averages, this helps us to overcome cross-section dependence caused by any economic crisis.

Equation (6) has confabulated the long-run coefficients from the short-run coefficients estimated by the CSARDL model.

$${\hat{\omega}}_{{\rm {csardl}},i} = \frac{{{\sum}_{K = 0}^{n_z} {\hat{\delta}_{K,i^{nw}}} }}{{1 - \sum_{K = 0}}}\widehat{\vartheta _{K,i}}$$
(6)

Next, we will derive the long-run coefficients and mean group estimates.

$${\widehat{\overline{\omega}}} _{{\rm {mg}}} = \frac{1}{n}\mathop {\sum}\limits_{i = 1}^n {{\widehat{\vartheta}}_i}$$
(7)

The short-run coefficients are

$$\begin{array}{ll}{\Delta}V_{i,t}=\sigma_{i}[V_{i,t-1}-\omega_{i}X_{i,t}]\mathop{\sum}\limits_{K=0}^{n_{w}}\vartheta_{K,i}, {\Delta} V_{i,t-1}\\ \qquad\qquad+\,\mathop{\sum}\limits_{K=0}^{n_{v}}\delta_{K,i}\Delta_{K}X_{i,t-1}+\mathop{\sum}\limits_{K=0}^{n_{y}} {\acute{\gamma}}_{i,},K{\bar{Y}}_{i,t-1}+\mu_{i.t}\end{array}$$
(8)

where \(\Delta _K = t - \left( {t - 1} \right)\)

Besides, short-run coefficients are

$${\hat{\tau}}_i = \left( {1 - \mathop {\sum}\limits_{K = 1}^{n_w} {\widehat{\vartheta_{K,i}}}} \right)$$
(9)
$${\hat{\omega}}_i = \frac{{{\sum}_{K = 0}^{n_v} {\hat{\delta }_{K,i}} }}{{\widehat{\tau_i}}}$$
(10)
$$\widehat{\bar{\omega}}_{{\rm {mg}}} = \frac{1}{n}\mathop{\sum}\limits_{i = 1}^n {\widehat {\vartheta _i}}$$
(11)

It is observed that the CSARDL approach is almost similar to the pooled mean group. In addition, the error correction term (Ect) elucidates the adjustment process towards equilibrium. The Ect quantifies the period an economy takes to reach equilibrium.

Robustness tests

In the presence of heterogeneity and Csd, the traditional econometrics approaches provide biased results (Yao et al., 2019). To get unbiased empirical results, we use two tests to check the robustness of the model; The first test is called augmented mean group (AGM) (Eberhardt and Teal, 2010). Similarly, the other test is established by Pesaran (2006), and it is called the common correlated effect mean group (CCEMG). These tests help researchers to cope with three issues, i.e., Csd, slope heterogeneity, and structural breaks. Another merit of applying the CCEMG is that it deals with the issue of identification and time-variant unobservable with heterogeneous slopes.

Furthermore, this test also overcomes all sorts of spillover effects by averaging dependent and independent variables for all the cross-sections. Simultaneously, it includes all kinds of global and country-specific shocks, such as oil price shocks or any other local spillover effects (Pesaran and Tosetti, 2011). The AMG is another compelling test considering issues like Csd, heterogeneity, and structural breaks. By including the years’ dummies, this test also solves the issues of unobservable factors (Eberhardt and Teal, 2010).

Empirical results and discussions

The present analysis confabulates the impact of Digi, Edu, Fert and GDP on FLFP for a sample of five developing economies, BRICS. Moreover, our analysis also illustrates the evidence of the “resource curse hypothesis” in these nations. In the past few years, economies have been economically and financially interconnected. Thus, fluctuation in aggregate demand of one economy may transfer to the other economies. In order to overcome the presence of any biasness and ambiguity in the model specification, we cannot ignore the integration of these financial, economic and cultural shocks within these economies. We may not get unbiased results if we do not consider these effects. Thus to overcome these issues, we apply the Csd test (Pesaran, 2015). We accepted the alternative hypothesis and inferred that our selected variables in selected countries are cross-sectionally dependent (Table 1).

By looking at issues like Csd, slope heterogeneity, and structural breaks, we utilized the two most suitable unit root tests belonging to the third-generation family. Table 2 shows the empirical outcomes of Pesaran (2007) and Bai and Carrion-I-Silvestre (2009) unit root tests. It is revealed from the empirical outcomes in Table 3 for Pesaran (2007) and Bai and Carrion-I-Silvestre (2009), which do not reject the null hypothesis at the level, i.e., I(0) and as illustrated previously that these tests have three main issues of panel data analysis. We do not reject the null hypothesis of non-stationarity for selected variables for the Bai and Carrion-I-Silvestre (2009) unit root test by considering the issue of possible structural breaks in the data. However, in the case of Pesaran (2007), all of the selected variables are stationary at the level. Considering this, we have utilized Bai and Carrion-I-Silvestre (2009) at the first difference, i.e., I(1). Table 3 also illustrates that by rejecting the null hypothesis, we know that selected variables, FLFP, Digi, Edu, Fert, and GDP are stationarity at the first difference.

Table 2 Unit root test with and without structural break Pesaran (2007).
Full size table
Table 3 Slope heterogeneity analysis.
Full size table

We applied Swamy’s slope homogeneity test (Pesaran and Yamagata, 2008). This test confabulates the existence of homogenous or heterogeneous slope coefficients. As per prior literature, homogeneous slope coefficients give misleading results (Zhong and Yang, 2022). Our empirical outcomes in Table 3 accept the alternative hypothesis, i.e. slopes are heterogeneous at a 1 per cent level of significance.

Once we find out about the order of integration, that is, I(1), we will specify the suitable methods for the long-run interconnection among the included variables. For this purpose, we will consider the cointegration test by Westerlund and Edgerton (2008). The null hypothesis is that there is no evidence of cointegration among the selected variables. The main advantage of this test is that we can use this test in the presence of the four most fundamental problems: heterogeneity, serial correlation, structural breaks, and cross-sectional dependence. The Westerlund and Edgerton (2008) cointegration estimation technique illustrates that all of the selected variables are cointegrated. Therefore, we can conclude a long-run interconnection among the selected variables for the sample of BRICS economies (Table 4).

Table 4 Westerlund and Edgerton (2008) panel cointegration analysis.
Full size table

We have elaborated on the estimated results of Banerjee and Carrion‐i‐Silvestre’s (2017) cointegration analysis in Table 5. The estimates confirm the cointegrating interconnections between FLFP, Edu, Fert and GDP and FLFP at a 1 per cent level of significance. This is also valid in the case of the full sample and for each country included in the sample, Brazil, Russia, India, China and South Africa. The empirical results for both tests, i.e. Westerlund and Edgerton (2008) and Banerjee and Carrion‐i‐Silvestre (2017), are consistent with the prior literature, for instance, Li et al. (2022) and Wei et al. (2022). Besides, the estimated results for Banerjee and Carrion‐i‐Silvestre (2017) for FLFP also support the cointegrating relationship.

Table 5 Banerjee and Carrion‐i‐Silvestre (2017) cointegration analysis.
Full size table

Table 6 illustrates the empirical outcomes of the CSARDL model. The empirical analysis confabulates that digitalization positively affects FLFP with a coefficient of 0.261, which is highly significant. It shows that a 1 per cent increase in access to the Internet is resulting in an increase in FLFP in the BRICS block. Hafeez et al. (2020) assess the role of ICT in improving FLFP in selected South Asian nations. As (ICT) is recognized as one of the significant factors in women’s empowerment. Ngoa and Song (2021) demonstrate that ICT plays a crucial role in encouraging FLFP in 48 African nations. According to Tüzemen et al. (2021), in the case of Turkey, there is not any connection between ICT and FLFP. Viollaz and Winkler, 2022 explain that in the case of Jordan, ICT contributes to the growth of FLFP because internet penetration increases online job search. Women with higher levels of education experience an increase in internet-based employment attainment. Galperin and Arcidiacono (2021) demonstrate that the correlation between employment and internet use is stronger among women than men in four Latin American countries. Asongu and Odhiambo (2022) prove that in the context of sub-Saharan African countries, ICT in the form of broadband subscriptions boosts the female employment rate.

Table 6 Results of CS-ARDL analysis (long run CS-ARDL results).
Full size table

Furthermore, females’ education is positively interconnected with FLFP for BRICS economies. It has a coefficient of 0.361 and is highly significant at a 1 per cent significance level. It means that a 1 per cent increase in females’ education leads to a 0.361 per cent expansion in FLFP. The findings are in line with the prior studies on the same idea. Onyeke and Ukwueze (2022) deduce in the context of Nigeria that the primary school education enrolment for females and secondary schooling enrolment rates for females have a positive long-term impact on the female labour force participation rate.

On the same line, the next included variable is fertility with a coefficient value of 0.220, and it has a negative interconnection with FLFP and is significant at 5 per cent. It reveals that a 1 per cent upsurge in fertility will lead to a 0.220 per cent decrease in FLFP. Our results align with the prior literature; for instance, Ukil (2015) states that fertility has a negative effect on FLFP because fertility is endogenous to female labour participation in the United Kingdom. Shittu and Abdullah (2019) explain that fertility hinders FLFP because it has a negative effect on FLFP in ASEAN-7 countries. Due to its close relationship with female labour force participation, Nazah et al. (2021) deduce that in 39 Asian nations, fertility had a detrimental effect on female labour force participation in the short term but not in the long term. Bawazir et al. (2022) exemplify that fertility hinders female labour force participation in Middle Eastern nations.

To investigate the impact of economic growth on FLFP, we have included GDP in our empirical model and elucidated its interconnection with FLFP. We have seen from the empirical outcomes of the CSARDL model that GDP is exerting a favourable influence on FLFP in these economies. It is also significant at 1 per cent. We have compared our inferences with the prior literature. Belke and Bolat (2016) conclude that the U-shaped relationship between economic growth and female labour participation is supported by the evidence from 148 developed and developing countries. Tasseven (2017) describes that the gross domestic product plays a crucial role in positively stimulating female labour force participation in G8 countries. Sasongko et al. (2020) demonstrate that economic growth has no impact on the female labour force participation rate in 34 provinces of Indonesia.

After illustrating the long-run empirical outcomes of the CARDL model, we will now elucidate the short-run results of the CSARDL model. We can infer from the empirical outcome shown in Table 7 that digitalization, education and GDP are positively interconnected with FLFP in BRICS economies in the short run. On the other hand, fertility is negatively and significantly associated with FLFP in the selected region. Moreover, the negative value of the error correction term highlighted convergence toward equilibrium, and the coefficient value of ECT is −0.211, which also reveals the peace of adjustment in the short run in these economies. In addition, the short-run coefficient values of Digi (−0.176), Edu (0.125), Fert (0.081), and GDP (0.071) are lower than their long-run magnitudes. It elucidated that these economies are developing economies.

Table 7 Results of CS-ARDL analysis (short-run CS-ARDL results).
Full size table
Table 8 Results of AMG and CCEMG for robustness check.
Full size table

The AMG and CCEMG tests are applied to assess the model’s robustness. Both tests elucidate the positive influence of digitalization on FLFP, and it is significant at 1 per cent. Furthermore, Edu and GDP also influence FLFP favourably, which is significant at 1 per cent. The fertility is also negatively connected with FLFP in AMG and CCEMG test estimates and is significant at 1 per cent. These results replicate our prior results from the CSARDL model for the five developing economies (Table 8).

Conclusions and policy implications

Even with the rapid and pervasive digital change of the global economy over the past decade, remarkably few studies have evaluated the effects of digitalization on labour market outcomes, particularly for women and emerging economies.

The prime motive of the present analysis is to discuss the impact of digitalization, education, fertility and GDP on FLFP in the panel of five BRICS economies. The study has utilized the latest available econometrics approaches to test these interconnections. For the empirical results, the analysis inspects the cross-section dependence using the latest available tests proposed by Pesaran (2015). After discovering the cross-section dependence, the present study applied the advanced, third-generation unit root tests. The empirical analysis also elucidated the presence of a long-run relationship among selected variables in the case of five developing economies.

Moreover, we have utilized the cross-sectional augmented autoregressive distributed lags (CSARDL) approach to validate the presence of short-run and long-run interconnections among these variables. The most significant contribution of this study is that it validates the existence of a positive connection between digitalization and FLFP in these economies. Additionally, gross domestic product is positively connected with FFLP; GDP also positively influences FLFP. However, the fertility rate is negatively associated with FLFP in the short and long run. Nonetheless, we have observed that the magnitude of the coefficients of these variables is higher in the long run compared to the short run. It shows that these economies are developing economies.

Our study extends a few significant policy insights based on our empirical estimated results. It is an essential quest for the policymakers in these economies to bring about policies based on promoting digitalization in these economies. These economies must focus on the new investment in information and communication technologies. To promote FLFP, it is important to give access to the Internet to the population at cheaper rates. The second most important factor is education. Women should be encouraged to get higher education in BRICS economies to promote FLFP. These economies need to make it mandatory for all women to get the higher secondary level of education. The policy upfront applies equally to the rest of the world, especially emerging nations. The current wave of literature and empirical evidence believe working from home has become normal. Therefore the nations must take aid from digitalization to run the labour markets smoothly. The empirical findings and implications open new horizons for developing nations to invest in digitization and education for females. The other blocks, for example, South Asian and MINT economies, can benefit from the present study’s findings.

Our empirical findings provide a few interesting insights for future research. In the future, researchers and academicians may utilize the same panel of the selected economies and validate the digitalization and FLFP interconnection by adding different macroeconomic variables, for example, inflation, poverty indices, fiscal and monetary policy tools, technological innovations and remittances. Future research could be focused on a different set of economies and regions, and most essentially, they may compare the macroeconomic policies of low, middle and high-income economies.