Nexus of Women’s Empowerment and Economic Growth in Saudi Arabia

Abstract

Saudi Arabia is actively boosting women’s labor force participation, exceeding 2022 targets to enhance human capital. The purpose of this study is to examine the nexus between women’s empowerment and economic growth using secondary data from 1997 to 2022. Empowerment was proxied by Saudi women employed in government, with Gross Domestic Product (GDP) and female B.Sc. holders used as explanatory variables. The result of the Johansen test depicts a long-run equilibrium relationship between these variables. The Vector Error Correction Model (VECM) revealed a significant negative short-term impact of GDP on women’s empowerment at the 5% level, while female graduates had a positive short-term effect. The model results indicated rapid adjustment, correcting about 71% of disequilibrium per period towards long-run equilibrium. Importantly, a significant positive long-run relationship exists between women’s empowerment and economic growth. Diagnostic tests confirmed the VECM’s reliability, reflected by normally distributed residuals, with no significant autocorrelation, and overall model stability. The study findings contribute valuable insights for policymakers and stakeholders striving to achieve gender equity and sustainable economic development. Moreover, qualitative methods could be employed in future research to enhance the comprehensive understanding of the cultural and social barriers to maximize the long-run virtuous cycle of empowerment and economic growth.

1. Introduction

1.1. Background

The Millennium Development Goals (MDGs) are critically evaluated by [1], revealing that the MDGs’ inadequate attention to social injustices hampered the advancement of women’s empowerment and gender equality. The emphasis on debt reduction and basic requirements overshadowed underlying gender inequality in many Southern countries, particularly in rural sub-Saharan Africa, where unequal access to land threatened the achievement of MDG 3.

The theoretical analysis of women’s empowerment on GDP and economic growth posits that gender inequality is a significant barrier to long-term economic prosperity, largely due to market inefficiencies and the underutilization of a nation’s human capital [2,3,4]. By limiting women’s access to education, health, and economic opportunities, a society fails to harness the full productive potential of half its population, which suppresses overall productivity, innovation, and economic resilience [2,5]. According to this framework, empowering women fosters growth through multiple channels. First, addressing gender gaps in human capital, such as education and health, leads to a more skilled and effective workforce [2]. Second, increasing women’s labor force participation and economic rights is seen as a direct way to boost a country’s GDP [4,5]. Furthermore, within the household, empowering women with control over resources, such as through employment or targeted cash transfers, shifts resource allocation toward investments in children’s education and health, leading to a more capable and productive next generation. This is based on the theory that women, who tend to value children’s well-being more than men, will maximize the “joint surplus” of human capital [6]. This also counters a “quantity-quality” trade-off in which limited opportunities for women lead to higher fertility and less investment per child [3]. Lastly, women’s political empowerment is theorized to be a driver of long-term economic growth by enhancing technological innovation and policy-making. By bringing diverse perspectives to the political arena, women can advocate for more effective policies that foster an environment conducive to entrepreneurship and technological change [7]. However, it is noted that the effects of empowerment are not automatic and can be limited by societal factors such as deeply entrenched social norms [3,8].

The concept of women’s empowerment in Saudi Arabia’s Vision 2030 has significant implications for women’s accomplishments. This is particularly evident in light of the substantial changes made to civil and labor laws in August 2019 that greatly enhanced women’s rights to employment, travel, and financial independence [9]. A significant long-term positive relationship between economic growth and women’s empowerment, measured by female employment in Saudi Arabia, was illustrated by [10], using time-series analysis that covered the period from 1999 to 2014.

As clearly indicated by the annual report on Saudi Arabia’s Vision 2030, significant strides have been made in economic diversification and private sector participation, which provide important cornerstones for the country’s long-term economic development [11,12]. Major accomplishments in 2022 include high-level growth of GDP, greater public health and urban infrastructure, and a better life for the people [13]. Vision 2030 catalyzed transformative reforms. Notable milestones comprised the removal of the driving prohibition in 2018, substantial legislative reforms that conferred upon women the right to travel, reside, and work without a male guardian’s consent, and promulgation of anti-harassment statutes [11,12,13,14,15]. Focused programs such as Qurrah (child-care subsidies) and Wusool (transportation subsidies) targeted pragmatic barriers [12,15]. Government directives, aimed at achieving 30% female total workforce participation, even affecting mandatory hiring in tourism/hospitality [15], resulted in cross-sector changes. The impact has been substantial. Women’s participation in the labor force increased upwards to 34.4–34.5% in 2022/2023, surpassing the targets, and they currently control 45% of SMEs [11,12]. By 2019, female tertiary enrollment was 70.9% [12]. Women are also more prominent in leadership, entrepreneurship, sports, arts, and non-traditional fields such as hospitality [11,12,14,15].

In 2016, before Vision 2030, there were strict systemic limitations on women in Saudi Arabia. Widespread male guardianship laws sharply restricted independence in travel, residence, and work [11,12,14,15,16]. There was a driving ban that severely restricted movement, and workforce participation for women was around 23.2% (2016), reflective of sociocultural biases. Educational paths were very limited, confining women to teaching, nursing, and at best, the medical field, not law, and they were just getting to the business [11,12,14,15]. Access to education, while increasing, only reached 25.2% female tertiary enrollment in 2000 [12,17,18], drawing attention to a surge in female labor supply in Saudi Arabia, which began circa 2019, following government initiatives for gender equality in the labor market. Before that, female unemployment was already high, and women were almost no part of the economy. The policy measures made great strides in job market access, education, economic participation, and gender equality, reducing the female unemployment rate to 5.7%.

Despite the ambitious empowerment goals of Saudi Vision 2030, research indicates the persistence of notable challenges within the labor market. While the population demonstrated nearly equal gender distribution in 2015, a significant labor market mismatch remains. This disparity is further evidenced by limited female representation in critical sectors such as engineering and business leadership, coupled with a continued male dominance in technical training programs and engineering admissions. Refs. [19,20] found that although the number of women attending college increased significantly between 1999 and 2013, women’s unemployment rates remained high. Even with consistently strong graduation rates, employment opportunities were predicted to be restricted for 2017–2020. However, equal opportunities for M.Sc. holders suggested that there was potential for more female employees with Ph.Ds. In addition to better matching graduate specializations with labor market demands, the research suggests expanding funding for female Ph.D. programs and systematically upskilling female professionals at the M.Sc. level. According to [21], Saudi Arabian women entrepreneurs have a poor grasp of sustainability, which limits the adoption of this concept as a profitable business prospect. The research highlighted the importance of improvement in sustainability education in order to increase awareness and encourage more significant social change among female entrepreneurs.

In spite of ambitious Saudi Vision 2030 targets to empower Saudi Arabian women, the contemporary literature tends to overlook the facts of women’s experiences [22]. The studies highlighted the enormous confluence of institutional and cultural determinants that could result in paradoxical, educational work and effort-related outcomes. They claimed that more fundamental factors such as social pressure, self-esteem, communication abilities, and career promotion need to be considered further to have a complete understanding of women’s empowerment in the Saudi Kingdom.

Despite progress, challenges persist post-reform. Supporting structures (50/100): World Bank baselines: women do not have access to affordable childcare, only partial implementation of parental leave, low level of inclusion in finance, and under movement towards senior management [11,12]. Social attitudes and cultural norms are still obstacles [11,14]. The policy reform agenda should be kept in place, and there is a continued demand for increased transition into supportive atmospheres as well as a change in conventions and society values to maintain development momentum beyond 2030 towards full economic participation and gender equilibrium [11,12,14,17]. According to [23], women’s employment involvement in Saudi Arabia is still far lower than men’s, despite the fact that the number of female graduates has greatly increased. One of the main recommendations is to overcome the sociocultural barriers that prevent women from fully participating in the workforce.

Women’s empowerment in Saudi Arabia, while a dynamic and evolving process, is increasingly being benchmarked against other GCC nations. Ref. [24] highlight a key distinction, noting that Saudi Arabia’s progress is largely driven by top-down, state-led reforms under Saudi Vision 2030, whereas other GCC countries have seen more influence from grassroots movements. This top-down approach is particularly focused on cracking the glass ceiling and increasing job opportunities for women in sectors like tourism and hospitality [25]. The analysis by [26] further benchmarks Saudi Arabia’s standing, noting that it ranks 127th on the Gender Gap Index, which is a significant gap behind the UAE’s 68th ranking. However, this still indicates notable progress for Saudi Arabia relative to others in the region.

Several studies identify common challenges and areas for improvement across the GCC. Refs. [27,28] both find that despite government support for women’s organizations and public sector employment, significant gaps remain in leadership positions due to sociocultural norms and discrimination in recruitment. Religion is a factor, but its influence is highly intertwined with historic and social factors [27]. Furthermore, women in the GCC face hurdles such as limited access to capital and a need to balance family responsibilities with their careers [24,26]. The role of women in entrepreneurship and on corporate boards is a key area of focus. Mohiuddin [29] spotlights the increasing role of women entrepreneurs in the GCC as they move beyond traditional roles to start their own businesses. Ellili [30] provides a financial perspective, showing that board gender diversity in the GCC has a significant, non-linear impact on corporate investment efficiency and sustainable growth. This suggests that increasing women’s presence on corporate boards is not just a social imperative but also a strategic business decision. The educational landscape, however, is a strong point, with women in all three GCC countries (Qatar, Saudi Arabia, and the UAE) outperforming boys in PISA test scores across all subjects, providing a solid foundation for their future contributions to the economy [26].

Ref. [31] study women’s empowerment perceptions in India across socioeconomic groups, contending that although laws to protect women and female participation are on the rise, gender inequality prevails. Based on the analysis of primary surveys, this study finds that women still have lower autonomy compared to men; however, their economic contribution enhances the security of the family, and attitudes towards empowerment become more favorable. Economic participation is one critical factor determining women’s empowerment in India, this study found.

At selected universities in China, Zhang [32] explores the challenges and progression of female leaders in organizations through a qualitative analysis of response activities, leadership principles, and development processes. Women are great decision-maker and inventors, but they still face so many challenges, such as balancing work, caring for family, and gender bias,” according to the survey”. To overcome biases and promote gender-equitable leadership, it highlights the importance of organizational climate, using inclusive leadership frameworks, and societal change.

The economic effect of women’s empowerment was explored by [33] by examining the appropriate mechanisms to increase female participation in the workforce. Drawing upon qualitative and quantitative methods, it examines existing levels of women’s economic participation, the extent, nature, and distribution of existing obstacles, and cost-effective measures to overcome them. The conclusion shows that investment in women is important for economic growth, social development, and human development. Ref. [34] studies women’s empowerment and economic development in India, suggesting that education and skills development are crucial for gender equality and sustainable development. The analysis focused on 317 women and five aspects of empowerment: economic, educational, social, civic, and family well-being. Indicating that formally educated tribal women (primary school and beyond) were not statistically significantly different from informally educated ones regarding empowerment but had higher scores in empowerment. The results underscore that investment in education and women’s career development should be sustained to ensure inclusive development and achieve India’s SDG-5 target. Education is an important dimension of both national development and women’s empowerment [35]. As society becomes more educated, women are more likely to assert their rights, promote social and economic development, and participate in the decision-making process. The investigation highlights the inequality that still exists in India regarding literacy and labor force participation rate, underscoring that education is critical for progress in gender equality and promoting inclusive development.

In Tehran, ref. [36] studied women’s psychological empowerment in connection to higher education and work tasks. A descriptive-practical analysis was conducted using Spearman correlation and the structural equations model. Women’s empowerment was the dependent variable, with work and education as the explanatory factors. Six hundred women in Tehran responded to the survey following random selection. This study’s conclusions demonstrated that administrative jobs and higher education positively impact women’s empowerment levels in Tehran. Ref. [37] argued that whereas transfers to women, particularly mothers, tend to increase spending on children and human capital investment, this does not always translate into broader economic development. Using a non-cooperative household model, they suggest that these transfers function best when human capital serves as the primary development engine. Data from Mexico’s PROGRESA program supports this, showing that payments to women raised spending on children and investments in human capital. In African higher education institutions, where significant positions are still held by men, Tsverukayi [38] draws attention to the persistent under-representation of women in senior leadership posts. The paper explores Black women’s experiences in policymaking positions at two Zimbabwean colleges using a feminist and decolonial framework, highlighting how decolonial women’s rights have influenced their leadership and empowerment strategies.

To achieve gender equality and empower women and girls, Ref. [39] conducts a bibliometric analysis of the Sustainable Development Goal (SDG) 5 research. The Web of Science database was used to find and examine 1095 relevant papers. The results show a growing interest in women’s empowerment worldwide in light of SDGs, with significant increases in research production noted in 2017 and 2021. Using keyword co-occurrence analysis, key themes such as job, education, autonomy, and power are highlighted. The paper presents a thorough analysis of academic developments in this field and makes suggestions to direct future investigations and influence legislative actions aimed at promoting women’s empowerment.

Based on the above background, this study is grounded in a body of research that links women’s empowerment and economic development, both globally and specifically within the Saudi Arabian context. The literature review establishes the theoretical and empirical foundation for this study’s purpose, research questions, and hypotheses. The Global Context and Theoretical Foundations indicate that the global efforts to link women’s empowerment with economic development, such as the Millennium Development Goals (MDGs), have been criticized for inadequately addressing social injustices and underlying gender inequality [1]. However, a growing body of the literature supports the idea that investing in women is crucial for economic growth and human development [33,34,35]. Studies from various countries highlight key dimensions of empowerment, including economic participation, education, and social autonomy [31,36]. Research also indicates that transfers to women can increase spending on human capital [37] and that women in leadership roles face challenges but are essential for promoting gender-equitable leadership [32,38]. This global perspective underscores that despite legislative progress, gender inequality persists and requires deliberate efforts to foster inclusive development.

This study’s focus on Saudi Arabia is particularly relevant due to the significant transformative reforms initiated under Vision 2030. These reforms, including legislative changes to civil and labor laws, have notably enhanced women’s rights to employment, travel, and financial independence [9,11,12,14,15]. These changes have led to a substantial increase in women’s labor force participation, exceeding initial targets [18]. A previous study by [10] already found a significant positive long-term relationship between economic growth and female employment in the Kingdom, providing an empirical basis for this study’s long-run hypothesis.

Despite this progress, the literature identifies persistent challenges. Research shows a mismatch between high female graduation rates and employment opportunities [19,20] and highlights the continued presence of socio-cultural barriers and institutional challenges like a lack of access to childcare and finance [11,12,23]. This body of literature emphasizes the need for a deeper understanding of the nexus between women’s empowerment and economic growth to ensure that reform momentum is sustained.

1.2. Purpose of This Study

This study is conducted to explore links between women’s empowerment and economic growth in Saudi Arabia during the period 1997–2022. The research is intended to study the interplay of short-run dynamics and long-term equilibrium relationships between women’s government-recorded employment, country GDP, and female tertiary education.

1.3. Study Questions

• Is there a long-run relationship among women’s government employment, GDP, and the number of female B.Sc. holders in Saudi Arabia?
• What is the short-term impact of GDP and female B.Sc. holders on women’s empowerment, as measured by government employment?
• How quickly does the relationship between women’s empowerment and economic growth adjust back to a long-run equilibrium after a shock?

1.4. Study Hypotheses

• H1 (Long-Run): There is a significant long-run positive relationship between women’s empowerment and economic growth in Saudi Arabia.
• H2 (Short-Run): Changes in GDP have a significant short-term impact on women’s empowerment.
• H3 (Short-Run): The number of female B.Sc. holders has a significant short-term positive impact on women’s empowerment.
• H4 (Equilibrium): The relationship between the variables has a stable and significant adjustment mechanism that corrects for short-term deviations and returns to a long-run equilibrium.

2. Materials and Methods

2.1. Data

This study employed a secondary macroeconomics dataset from 1997 to 2023 with the following variables: number of Saudi employed women in the government sector (safemG) as a proxy for women’s empowerment. The data were acquired from the Saudi Central Bank Statistical Report 2025 [40]. This study used employed women in the government sector due to the unavailability of time series data for women’s employment in the private sector, as it covers the period from 2005 to 2022. The data for Gross Domestic Product (GDP) (billions of USD) were obtained from Macrotrends [41], whereas the number of female B.Sc. holders (univfem) was collected from the Saudi Central Bank [40].

The argument that government employment serves as a more effective proxy for women’s empowerment than private sector jobs is substantiated by several key factors. First, government positions typically offer greater job security, stable salaries, and benefits, enabling women to make informed choices about education and family planning [42]. Additionally, public sector jobs often have established policies promoting gender equity, such as maternity leave and anti-discrimination laws, creating environments conducive to women’s advancement [43]. The clarity of career development pathways in government roles empowers women to pursue leadership positions, thereby influencing policy and decision-making [7]. Furthermore, women in government can advocate for inclusive policies, enhancing their representation and influence in societal matters. Social norms also play a role; public sector jobs are often viewed as prestigious, encouraging more women to enter these fields [42]. Lastly, government employees frequently engage with community programs, allowing them to witness the direct impact of their work [42]. In contrast, private sector jobs often lack the supportive frameworks necessary for women’s empowerment, thus reinforcing the argument for government employment as a more reliable measure of women’s agency.

Figure 1 shows that the trend of Saudi employed women in both the government and private sectors during the period from 2005 to 2022 exhibits an upward trend, with the number of Saudi women employed in the private sector (49 thousand) exceeding that in the government sector (19 thousand) each year. Figure 2 illustrates an increasing trend in female B.Sc. holders during the period 1997–2022, with the year 2020 witnessing the highest number of female B.Sc. graduates. In the same line, it shows an increasing trend in GDP during the same period, reflecting an increasing trend in the economic growth of the country.

2.2. Method

This section outlines how this study’s objectives and hypotheses were addressed using specific data sources and statistical methods. The methodology included unit root tests to verify the stationarity of variables. Lag selection was performed to determine the optimal lag length for the model. Johansen Cointegration Tests established long-run equilibrium relationships among the variables. Ultimately, the Vector Error Correction Model (VECM) was used to analyze both short-run dynamics and long-term relationships while accounting for endogeneity.

• Unit Root Test

Persistence tests (unit root tests) are a fundamental requirement for meaningful time series econometric analyses, such as using models like the Vector Error Correction Model (VECM) to investigate complex and lasting interactions like the empowerment of women. The main purpose of these tests is to establish whether a time series has constant statistical properties (mean, variance, or autocorrelation over time) (i.e., it is stationary) or exhibits trends or other forms of structure that indicate non-stationarity, such as the presence of a unit root [44,45,46,47,48]. This comes with a noble aim, currently crucial, because often when standard regression techniques (such as OLS) or VECM are applied to non-stationary series, a vast majority of the time, spurious relations and misleading inferences are generated, and any conclusions about causal or long-run relationships are discarded [44,46,48].

Ensuring a stationary series has direct implications for model form and predictive accuracy. Data must be differenced to achieve stationarity using unit root tests (ADF, KPSS, PP) since non-stationarity has been confirmed, and only then can VECM be estimated [44,45,48]. These tests are, moreover, necessary to detect cointegration. When non-stationary variables (integrated of the same order, i.e., I (1)) are cointegrated, VECM becomes the appropriate model to study both short-run dynamics and long-run adjustments, which is essential to address persistent evidence as in the case of determinants of women’s empowerment [44,46,48].

• Lag Selection

The most important issue in a VECM is the proper lag length, as it basically affects how well the model can significantly the dynamic relationship between the integrated time series variables [49,50]. Its primary aim is to trade off model complexity against sufficiency by including enough lags to capture lagged influences on current behavior and equilibrium adjustments but not adding additional parameters that weaken robustness [51,52]. As a result, appropriate lag selection is crucial to obtain valid inference, trusted forecasts, and the discovery of real economic relationships [49,50]. Structurally, the number of lags in VECM (p-1) is determined directly by the number of lags of the VAR model in levels (p lags), emphasizing again its roots in cointegration analysis, where the error correction term determines the long-run equilibrium convergence [52]. Finally, severe lag selection ensures that the VECM reliably traces both short-run fluctuations and long-run cointegrating links, a necessity for interpreting policy insights in dynamic systems such as women’s empowerment and economic growth [50,52].

• Johansen Tests for Cointegration

These Tests are essential for studying long-run relationships between non-stationary time series in the context of the Vector Error Correction Model (VECM). These tests provide identification of cointegrating relationships characterizing the simultaneous short-term dynamics and long-run equilibrium relationships. By determining the extent to which economic variables (for example, income and consumption) move together over time, the test helps prevent spurious regression, thus improving the validity of estimates [53,54]. Its methodological robustness of estimating the number of cointegrating vectors is particularly useful for grasping intricate interrelations in economic systems and enabling better policy making and forecasting [50,55,56,57]. Second, the Johansen approach allows for structural breaks, which provide more robustness in empirical investigations. By and large, including these tests in VECM analysis is justified, as it enables the exploration of dynamic relationships, contributing to more robust econometric models and better-informed decisions in all areas of human activity, including financial and economic ones [54].

• The Vector Error Correction Model (VECM)

This Test is a critical method for modeling the dynamic interrelationship among variables, as it accurately depicts the short-term dynamics and long-run equilibrium relationships across non-stationary time series variables. The contribution of VECM is in its ability to determine cointegration among the variables, allowing for the observation of significant links that are helpful in the policy formulation to achieve the goal of gender equality [58,59]. The model reduces the possibility of spurious regression, enabling the robust examination of how empowering women promotes sustainable economic development [50,54]. Furthermore, the VECM can control for structural breaks and endogeneity too and has developed into a useful tool to investigate complex economic dynamics [55,60]. The propensity to engage in this kind of discriminatory behavior is not unique to Pakistan and may have broader applicability in the formulation of targeted policies to increase the role of women in the economy and thereby the expected economic payoffs [57,61]. Overall, the VECM not only facilitates effective analysis of the nexus between variables but also supports evidence-based policymaking and strategic planning to foster inclusive growth across various sectors [56].

According to [62], the autoregressive modeling category, features three key approaches, one of which is VAR (Vector Autoregression), which models dynamic interrelationships among multiple time series by regressing each variable on its own lags and the lags of others in a multi-equation framework; however, it does not account for long-term equilibrium relationships. VECM (Vector Error Correction Model) addresses this limitation by explicitly modeling long-term equilibrium through an error correction term, requiring variables to be cointegrated (sharing a stable long-run relationship) and integrated of the same order. In contrast, ARDL (Autoregressive Distributed Lag) models analyze relationships within a single equation, uniquely accommodating both stationary and non-stationary variables of different integration orders without requiring cointegration, capturing short- and long-term dynamics using lagged variables and their differences.

Accordingly, in this study, the VECM model is elucidated in Equations (1) and (2) as follows:

$$∆ln{\mathrm{s}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{G}}_t={\beta }_0+{\sum }_{i=1}^n{\alpha }_1∆ln{\mathrm{s}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{G}}_{t-i}+{\sum }_{j=1}^n{\alpha }_2∆{GDP}_{t-i}+{\sum }_{j=1}^n{\alpha }_3∆ln{\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{v}\mathrm{f}\mathrm{e}\mathrm{m}}_{t-i}+\lambda EC{T}_{t-1}+e$$

(1)

$$\lambda EC{T}_{t-1}=ln{\mathrm{s}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{G}}_{t-i}-{\beta }_0-{\beta }_1{GDP}_{t-i}-{\beta }_{2}ln{F\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{v}\mathrm{f}\mathrm{e}\mathrm{m}}_{t-i}$$

(2)

where

safemG represents the number of Saudi women employed in the government sector, GDP = Gross Domestic Product, univfem = number of female B.Sc. holders, ${\beta }_0$ = intercept, $\mathsf{\Delta }$ = the first difference, i = number of lags, ${\beta }_1,{\beta }_2$ = long-run coefficients, ${\alpha }_1,{\alpha }_2,{\alpha }_3$ = short-run coefficients, $\lambda$ = speed of adjustment, and $EC{T}_{t-1}$ = error correction term.

3. Result and Discussion

• Descriptive statistics

Descriptive statistics and normality tests are foundational statistics that are not only descriptive but also hold significant interpretative value for understanding the dynamics of women’s empowerment within the Vector Error Correction Model (VECM) framework.

Table 1. Descriptive statistics of the study variables.

1. Summary	safemG	GDP	GDP
Mean	336,762.1	519.3	70,956.3
Std. Dev.	124,224.5	279.9	42,212.4
Min	181,653.0	146.8	21,221.0
Max	502,578.0	1108.6	194,784.0
2. Normality Test
a.Skewness and kurtosis tests
Statistics test	Pr (Skewness)	Pr (Kurtosis)	adj chi2 (2)	Prob > chi2
safemG	0.4993	0.000	16.8	0.0002
GDP	0.7283	0.0672	3.78	0.1508
univfem	0.0086	0.1199	8.00	0.0184
b. Shapiro–Wilk
Statistics test	W	V	z	Prob > z
safemG	0.83158	4.816	3.221	0.00064
GDP	0.92095	2.261	1.671	0.04732
univfem	0.87308	3.629	2.642	0.00413

summarizes the central tendency and dispersion of the study variables over the period 1997–2022. The average number of Saudi women employed in government (safemG) is 336,762, GDP averages USD 519.3 billion, and there are an average of 70,956 female B.Sc. holders. However, the substantial standard deviations (safemG: 124,224.5, GDP: 279.9, and univfem: 42,212.4) indicate considerable variability over the 26-year period. The reported minimum and maximum values further illustrate the range of observed data. Research underscores the crucial role of descriptive statistics in VECM-based women’s empowerment analysis. These statistics provide the necessary context for interpreting the model’s results by summarizing trends, patterns, and the magnitude of variation in the underlying data [62,63,64,65,66]. The averages establish baseline levels of key empowerment indicators, while the high variability, particularly for safemG, suggests significant changes or uneven progress in women’s government employment over time, which is vital for understanding long-run equilibrium relationships within the VECM.

The normality tests (skewness, kurtosis, and Shapiro–Wilk) presented in Table 1 indicate deviations from a normal distribution for most variables, as Pr (skewness) and Pr (kurtosis) values close to 0.05 or lower, typically suggest that the data are not normally distributed at the 5% level of significance, especially for SafemG and Univfem, and this also demonstrates clear non-normality based on skewness. Hence, to overcome the non-normality problem, they are specified in their natural logarithm forms. In contrast to the GDP, it exhibits normal and borderline characteristics (Shapiro–Wilk Prob > z = 0.04732, Prob > chi2 = 0.1508), and, therefore, it is used in the model in its current form.

Understanding the distributional properties, as revealed by these tests, remains essential for correctly specifying and interpreting the VECM, even if the model itself is less sensitive to this assumption than some others. It informs researchers about the data’s characteristics and potential complexities [63].

• Unit Root Test (Stationary Test)

The Augmented Dickey–Fuller (ADF) presented in

Table 2. Unit root test (stationary test).

Variable	Levels (0)		First Difference (1)		Outcome
	T-Statistic	Critical (5%)	T-Statistic	Critical (5%)
lnsafemG	−0.686	−2.518	−2.584	−2.528	I (1)
GDP	0.124	−3.75	−4.328	−3.75	I (1)
lnunivfem	−1.162	−1.162	−3.722	−2.528	I (1)

, revealed that the variables lnsafemG, GDP, and lnunivfem were all found to be non-stationary in their levels, as the calculated T-Statistic is less than the critical values at the 5% level of significance; hence, we fail to reject the unit root null hypothesis. However, after first differencing (I (1)), they achieved stationarity (T-Statistics: −2.584, −4.328, −3.889, and −3.722), confirming their classification as I (1) processes. It is worth mentioning that without rigorous stationary testing, this study’s conclusions regarding this nexus would lack a sound econometric foundation, ensuring that the results are robust and reliable and in line with [44,48].

• Lag Selection

This study applied commonly used techniques to determine the appropriate lag length encompassing primarily Akaike (AIC), Schwarz/Bayesian (BIC/SC), and Hannan–Quinn (HQC/HQIC) to objectively minimize information loss and balance fit against frugality in line with [49,51]. Ultimately, the chosen model appears to balance these criteria, likely using lag 4 for the subsequent VECM, as it contains lower values for most criteria, denoted by star sign (*) (

Table 3. Lag selection.

lag	LL	LR	df	p	FPE	AIC	HQIC	SBIC
0	−133.4				48.609	12.3973	12.4323	12.546
1	−70.02	126.7	9	0.00	0.351693	7.45637	7.59656	8.05148
2	−55.44	29.162	9	0.001	0.22403	6.94901	7.19434	7.99046 *
3	−42.14	26.604	9	0.002	0.175435 *	6.55793	6.90841	8.04572
4	−31.77	20.739 *	9	0.014	0.211977	6.43344 *	6.88906 *	8.36756

).

• Analysis of Johansen Tests for Cointegration

The Johansen tests assess whether a long-run equilibrium relationship, or “cointegration,” exists among the variables. The results (

Table 4. Johansen tests for cointegration.

Maximum Rank	Parms	LL	Eigenvalue	Trace Statistic	Critical Value (5%)
0	30	−52.3612		41.1867	29.68
1	35	−38.715	0.71078	13.8943 *	15.41
2	38	−31.9158	0.46104	0.2959	3.76
3	39	−31.7678	0.01336
Maximum Rank	Parms	LL	Eigenvalue	Max Statistic	Critical Value (5%)
0	30	−52.3612		27.2924	20.97
1	35	−38.715	0.71078	13.5984	14.07
2	38	−31.9158	0.46104	0.2959	3.76
3	39	−31.7678	0.01336

) reveal important findings regarding the presence and number of cointegrating relationships. In the Trace statistic test, the test for maximum rank 0, which posits no cointegrating relationships, yields a trace statistic of 41.2, exceeding the critical value of 29.7 at the 5% level. This allows us to reject the null hypothesis of zero cointegration.

For maximum rank 1, the trace statistic of 13.9 (denoted by *) is less than the critical value of 15.4, so we do not reject the hypothesis of at most one cointegrating relationship. The result from the Trace test then indicates that there is one cointegrating relationship.

Also, in the maximum Eigenvalue statistic test, for the test of maximum rank 0, the maximum statistic is 27.3, which is higher than the critical value of 20.97, enabling us to reject the hypothesis of no cointegration. In contrast, the test for maximum rank 1 implies a maximum statistic of 13.6, which is less than the critical value of 14.07, and thus, we do not reject the null of at most one cointegrating vector. This also implies one cointegrating vector.

Overall, both tests consistently indicate that there is one cointegrating relationship among the variables (lnsafemG, lnGDP, and lnunivfem). This important result demonstrates a sustainable long-run equilibrium link among the study variables and provides support for applying a vector error-correction model (VECM) to investigate their dynamic co-movements.

• Analysis of the Vector Error-Correction Model (VECM)

The summary of the results for women’s empowerment (lnsafemG equation) illustrated in

Table 5. Vector error-correction model.

Equation	Parms	RMSE	R-sq	chi2	p > chi2
D_lnsafemG	11	0.030951	0.8807	81.24127	0.000
D_GDP	11	70.9345	0.6948	25.04403	0.009
D_lnunivfem	11	0.129848	0.7661	36.02939	0.0002
	Coef.	Std. Err.	z	p > z	[95% Conf. Interval]
D_lnsafemG
_ce1 L1.	−0.71002	0.172691	−4.11	0.000	−1.04848	−0.37155
lnsafemG
LD.	0.51786	0.194638	2.66	0.008	0.136376	0.899344
L2D.	0.089742	0.171939	0.52	0.602	−0.24725	0.426735
L3D.	−0.12516	0.16927	−0.74	0.46	−0.45692	0.206608
GDP
LD.	−0.00064	0.000248	−2.56	0.01	−0.00112	−0.00015
L2D.	−0.00034	0.000194	−1.76	0.078	−0.00072	0.0000382
L3D.	−0.00058	0.000184	−3.16	0.002	−0.00094	−0.00022
lnunivfem
LD.	0.12063	0.064473	1.87	0.061	−0.00573	0.246994
L2D.	0.071292	0.110512	0.65	0.519	−0.14531	0.287892
L3D.	0.176033	0.118813	1.48	0.138	−0.05684	0.408902
_cons	−0.43961	0.119466	−3.68	0	−0.67376	−0.20546
D_GDP
_ce1
L1.	−240.709	395.7758	−0.61	0.543	−1016.42	534.9969
lnsafemG
LD.	195.0193	446.0759	0.44	0.662	−679.274	1069.312
L2D.	−187.273	394.0525	−0.48	0.635	−959.601	585.0562
L3D.	−853.254	387.9363	−2.2	0.028	−1613.6	−92.9128
GDP
LD.	−0.54201	0.568459	−0.95	0.34	−1.65617	0.572154
L2D.	−0.88374	0.444161	−1.99	0.047	−1.75428	−0.0132
L3D.	−0.35523	0.421171	−0.84	0.399	−1.18071	0.470248
lnunivfem
LD.	−39.4693	147.7598	−0.27	0.789	−329.073	250.1346
L2D.	7.519649	253.2732	0.03	0.976	−488.887	503.926
L3D.	−354.683	272.298	−1.3	0.193	−888.377	179.0113
_cons	0.004542	273.7936	0	1	−536.621	536.6302
D_lnunivfem
_ce1
L1.	1.023355	0.724481	1.41	0.158	−0.3966	2.443311
lnsafemG
LD.	−0.27149	0.816557	−0.33	0.74	−1.87191	1.328931
L2D.	0.24108	0.721326	0.33	0.738	−1.17269	1.654853
L3D.	1.807823	0.71013	2.55	0.011	0.415994	3.199652
GDP
LD.	0.00067	0.001041	0.64	0.52	−0.00137	0.00271
L2D.	0.000892	0.000813	1.1	0.273	−0.0007	0.002486
L3D.	−0.0003	0.000771	−0.39	0.698	−0.00181	0.001212
lnunivfem
LD.	−1.1297	0.270479	−4.18	0	−1.65983	−0.59957
L2D.	−0.7591	0.463625	−1.64	0.102	−1.66779	0.149589
L3D.	0.249468	0.49845	0.5	0.617	−0.72748	1.226413
_cons	0.763282	0.501188	1.52	0.128	−0.21903	1.745593
Identification: beta is exactly identified.
Normalized co-integrating equation
Beta	Coef.	Std. Err.	z	p > z	[95% Conf. Interval]
_ce1
lnsafemG	1
GDP	−0.0014	0.000185	−7.55	0.000	−0.00176	−0.00103
lnunivfem	−0.02257	0.095015	−0.24	0.812	−0.20879	0.16366
_cons	−12.3761

reveals several key metrics: the model parameters indicate a good fit, with R-squared values of 0.8807, indicating that 88% of the variation in women’s empowerment is explained by GDP and the number of female B.Sc. holders. In addition, p > chi2 values were well below 0.05, confirming that the variables jointly explain changes in the dependent variables.

The error correction term (_ce1) is pivotal, reflecting how quickly the system returns to equilibrium after a shock. The coefficient of −0.71002 is highly significant, indicating that approximately 71% of disequilibrium is corrected in the current period, suggesting a rapid adjustment towards equilibrium.

Short-run coefficients show immediate impacts of changes in variables. A past increase in female government employment (lagged) significantly boosts current employment changes at the 5% level of significance, while an increase in GDP (lagged one and three) has significant negative impacts on women’s empowerment. The justification of this negative short-term GDP impact can be attributed to diverse drivers. For instance, research shows that when women enter traditionally male-dominated industries, companies may incur short-term expenses related to training and infrastructure development. Ref. [67] note that such integration necessitates organizational changes and training, which can result in upfront costs before realizing productivity gains. Additionally, the transition process can be disruptive in sectors heavily dependent on male labor, as highlighted by [68], indicating that firms often face inefficiencies while adjusting management styles and workflows to accommodate a more diverse workforce, ultimately reducing productivity during this adaptation phase. The concept of “labor market frictions,” discussed by [69], further supports this view, as reallocating labor can induce short-term inefficiencies due to uncertainty or resistance from existing employees, resulting in decreased productivity. Moreover, ref. [70] reveals that short-term investments in women’s empowerment can redirect funds from traditional growth-enhancing projects, adversely affecting overall economic performance before the benefits of such initiatives materialize. Furthermore, ref. [71] highlight how cultural norms and societal resistance can hinder workplace integration, creating friction that delays productivity gains and economic growth. Collectively, these findings provide robust evidence that incorporating more women into the workforce may initially negatively impact GDP, illustrating the intricate relationship between social change and economic growth and suggesting that further research is needed to identify ways to mitigate these challenges and maximize the long-term benefits of gender integration across various sectors. In addition, the study results reveal a positive short-run relationship between women’s empowerment and the number of female B.Sc. holders at a 10% level of significance.

The Johansen-normalized cointegrating equation reveals a positive significant long-run relationship between women’s empowerment and economic growth, with a one-unit increase in GDP associated with a 0.0014 increase in female government employment in the long run. The intercept of this long-run relationship is the constant term (12.3761). Overall, these findings confirm this dynamic relationship between the variables and the applicability of the VECM in capturing such a relationship.

This study also suggests upholding the Vision 2030 policies in increasing women’s labor force participation to offset both the short-run negative effect on GDP and to capitalize on the long-run virtuous cycle of empowerment and economic growth.

4. Diagnostic Tests

Diagnostic tests are important for confirming that the VECM assumptions hold, and consequently the model is valid. The autocorrelation test (Lagrange-multiplier test) (

Table 6. Lagrange-multiplier test.

Lag	chi2	df	Prob > chi2
1	6.2365	9	0.71603
2	12.9577	9	0.16453

) examines whether the residuals from the model are correlated over time. Prob > chi2 is much higher than 0.05. This means we cannot reject the null hypothesis of no autocorrelation, indicating that our model adequately accounts for the dynamic relationships and no significant autocorrelation is present in the residuals.

The normality tests (Jarque-Bera, skewness, and kurtosis) are reported in

Table 7. Normality test using Jarque-Bera, skewness, and kurtosis tests.

Jarque-Bera Test
Equation	chi2	df	Prob > chi2
D_lnsafemG	3.4	2	0.18264
D_GDP	0.026	2	0.98693
D_lnunivfem	0.464	2	0.79293
ALL	3.891	6	0.69144
Skewness Test
Equation	Skewness	chi2	df	Prob > chi2
D_lnsafemG	−0.83592	2.562	1	0.10945
D_GDP	−0.08473	0.026	1	0.87112
D_lnunivfem	0.29055	0.31	1	0.57796
ALL		2.898	3	0.40762
Kurtosis Test
Equation	Kurtosis	chi2	df	Prob > chi2
D_lnsafemG	3.9563	0.838	1	0.35986
D_GDP	2.9999	0	1	0.9999
D_lnunivfem	2.5895	0.155	1	0.69427
ALL		0.993	3	0.80298

to test whether the residuals are normally distributed. The Jarque-Bera test returns a Prob > chi2 of 0.69144 (which is significantly larger than 0.05); thus, we do not reject the null that the series in X is normally distributed. Likewise, the skewness test has a Prob > chi2 of 0.40762 (not significantly skewed), and the kurtosis test has a Prob > chi2 of 0.80298 (not significantly kurtotic). Overall, these tests provide evidence that the VECM residuals are normally distributed, which is desirable for model fit.

The stability test (

Table 8. Stability test.

Eigenvalue		Modulus
−0.5946273 +	0.9772123i	1.14391
−0.5946273 −	0.9772123i	1.14391
1		1
1		1
0.8394186 +	0.4289706i	0.942677
0.8394186 −	0.4289706i	0.942677
0.325564 +	0.8783976i	0.936789
0.325564 −	0.8783976i	0.936789
−0.8421371 +	0.07547812i	0.845513
−0.8421371 −	0.07547812i	0.845513
−0.00377845 +	0.768783i	0.768792
−0.00377845 −	0.768783i	0.768792

The VECM specification imposes 2 unit moduli.

) ensures that the model’s dynamics are stable and that disturbances do not give rise to explosive dynamics. For stability, the absolute value of all eigenvalues (except those associated with cointegrating vectors, for which the value is 1) is less than 1. The eigenvalues represent two values with absolute value 1 that must be the cointegrating relationship. All but one of the eigenvalues are of modulus lower than 1 (e.g., 0.942677, 0.936789, 0.845513, and 0.768792). This means the VECM is stable, implying well-behaved dynamics in the model, and deviations to the LRE are converging back to it.

5. Conclusions

This study explores the connection between women’s empowerment and economic growth in Saudi Arabia from 1997 to 2022, emphasizing the role of government employment as a proxy for women’s empowerment. The Johansen test confirms the presence of a cointegrating relationship among the variables. The study findings reveal a significant long-term positive relationship between women’s empowerment, indicated by government employment, and economic growth, while the Vector Error Correction Model (VECM) highlights that approximately 71% of any disequilibrium adjusts back to equilibrium in subsequent periods.

Despite the progress made under Saudi Vision 2030, which has catalyzed substantial reforms for women’s rights and labor participation, this study identifies persistent challenges. Notably, the short-term negative impact of GDP growth on women’s empowerment suggests the need for complementary economic policies to mitigate these effects. Furthermore, this study reveals that female B.Sc. holders positively influence women’s empowerment.

The implications of this research underscore the importance of maintaining and enhancing initiatives aimed at increasing women’s labor force participation, thereby fostering a virtuous cycle of empowerment and economic growth. Future research should address the limitations of this study, particularly the exclusion of private sector dynamics, to provide a better understanding of women’s empowerment in the Saudi economic landscape. Overall, the findings contribute valuable insights for policymakers and stakeholders striving to achieve gender equity and sustainable economic development in Saudi Arabia. Moreover, qualitative methods, such as surveys and interviews, could be employed in future research to enhance the understanding of the cultural and social barriers affecting the subject matter, offering a more comprehensive perspective.