Women in Transition: The Dynamic Effects of Inward FDI on Female Employment in the Economy and Across Sectors

Abstract

This paper examines the effects of inward Foreign Direct Investment (FDI) on the female employment rate in the economy and the share of female employment across sectors. The empirical analysis is implemented through the Generalized Method of Moments (GMM) System estimator for dynamic panel models using different empirical specifications and FDI openness indicators. The main results show that the overall effects of inward FDI on the national female employment rate are not statistically significant. However, they reveal that inward FDI has promoted the share of female employment in the service sector and has led to decreases in the share of female employment in agriculture. The FDI effects on the share of female employment in the industrial sector are found to be statistically insignificant. These results are generally supported when running the empirical analysis through alternative FDI openness indicators. Also, supplementary analysis reveals some variations in the magnitude of these effects over different national income categories. The findings in this paper emphasize FDI’s gendered influences in the labour market. They are consistent with the prevalence of macroeconomic channels through which inward FDI impacts female employment across sectors, and they encompass the underlying implications of various counteracting microeconomic factors.

1. Introduction

The relationship between economic globalization and gender inequality in the labour market is a complex theme, and it has been regularly debated in policy forums and analyzed in the academic literature. The favourable standpoints on global economic integration for reducing gender inequality in the labour market are mainly associated with the positive effects of economic growth, improved social institutions, and resulting structural changes in women’s economic and educational opportunities and social status (Gray et al. 2006; The World Bank 2011; Potrafke and Ursprung 2012). In contrast, some critical perspectives on the implications of economic globalization for women cast concerns over the deterioration in the working conditions and the pervasiveness of the gender wage gap, particularly in exporting labour-intensive industries (Wood 1997; Blecker and Seguino 2002; Baliamoune-Lutz 2007), and over the diffusion of advanced technologies that disproportionally displace female workers who constitute larger shares of employment in several labour-intensive industries (Tejani and Milberg 2016). Despite these diverging reviews, there is a general emphasis that economic globalization has led to a reconfiguration of international economic activities (McMillan and Rodrik 2011; Wood 2019) and has fostered gradual shifts in national economies from agriculture toward manufacturing and service (Mehra and Gammage 1999; Barrientos et al. 2004). Accordingly, there have been significant dynamic implications for the labour market in terms of employment characteristics and composition across countries and sectors.

One of the main characteristics of economic globalization is the spectacular surge in Foreign Direct Investment (FDI) and the activities of Multinational Enterprises (MNEs). Inward FDI often generates positive macroeconomic effects on capital formation and economic growth (Borensztein et al. 1998; Alfaro et al. 2004), and fosters various microeconomic conduits in benefiting from under-utilized female labour (Siegmann 2006; Oostendorp 2009; Standing 2010), transferring new business culture (Lawler and Bae 1998; Tang and Zhang 2021), assigning capital-intensive industrial processes and advanced technology (Harrison and Rodríguez-Clare 2010; Tejani and Milberg 2016; Pham et al. 2021), and increasing market competition (Chen et al. 2013; Heyman et al. 2013; Vahter and Masso 2019).1 Accordingly, inward FDI generates significant effects on women’s economic status and employment in the economy and across sectors. However, increases in female employment may not necessarily reflect improvement in women’s conditions in the workplace and labour market. For instance, Standing (1989) notes that economic globalization has led to significant growth in production accompanied by increases in female employment. However, Standing (1989) also underlines that the adoption of flexible labour configurations was followed by the deterioration in the working conditions of female workers. This point was further emphasized by Moghadam (1999), who highlights the drawbacks that economic globalization inflicts on female working conditions.

Foreign affiliates of MNEs are commonly portrayed to have higher propensities to employ female workers compared to domestic firms in host countries. This characterization is supported by firm-level statistics and empirical evidence that underscore higher female employment rates in foreign affiliates of MNEs in developed and developing countries (e.g., Chen et al. 2013; Fakih and Ghazalian 2015; Kodama et al. 2018; Siegel et al. 2019).2 Firm-level empirical studies significantly contribute to comprehending the microeconomic and business mechanisms through which foreign affiliates respond to female employment. However, they do not encompass the macroeconomic effects of inward FDI on female employment in the economy and the dynamics of female employment transition across sectors. Furthermore, firm-level empirical analyses are typically conducted for specific countries, regions, and industries. Hence, they do not stretch to cover information on international variations in inward FDI, national economic and socio-economic characteristics, and female employment rates.

Despite the growing attention on the influence of FDI on labour markets, limited research has specifically explored its dynamic gendered influences and sectoral variations in employment outcomes. To address this limitation, this paper examines the effects of inward FDI on female employment in the economy and across sectors using a panel dataset that spans the time period 2001–2019 and covers 148 countries. The empirical analysis accounts for the overall implications of various macroeconomic and microeconomic channels through which inward FDI affects female employment, and it examines the distinct effects across sectors. The empirical analysis is implemented through the Generalized Method of Moments (GMM) System estimator for dynamic panel models using different empirical specifications, and it estimates the short-run and the long-run effects of inward FDI on the female employment rate in the economy and on the share of female employment across sectors. Also, the supplementary contribution of this paper is carried through the empirical methodology that controls for the confounding factors and through the use of alternative FDI openness indicators that depict the extent of foreign investment openness, capital operation freedom, and freedom of foreigners to visit.

The remainder of this paper is structured as follows. Section 2 overviews the macroeconomic and microeconomic channels through which inward FDI affects female employment in the economy and across sectors. Section 3 presents the empirical model and econometric methodology, and it overviews the data and variables. Section 4 presents and discusses the benchmark empirical results, and Section 5 implements robustness checks by executing the empirical analysis using alternative FDI openness variables and empirical specifications. Section 6 provides concluding remarks.

2. How Does Inward FDI Affect Female Employment?

FDI is often characterized as a macroeconomic vector that accelerates economic growth and fosters economic development (Borensztein et al. 1998; Alfaro et al. 2004; Iamsiraroj 2016).3 Economic growth eventually leads to increases in national income, and it generally induces structural economic changes that broaden women’s employment opportunities and induce the transition of female workers from the agricultural sector toward the (expanding) industrial and service sectors (Boserup 1970; Goldin 1995; Tam 2011; Ghazalian 2022). Hence, the catalyst role of FDI in stimulating economic growth constitutes an indirect channel through which FDI inflows decrease the shares of female employment in the agricultural sector and raise the share of female employment in the industrial and service sectors. Moreover, higher levels of national income are normally accompanied by increases in school enrolment of women, improvements in social infrastructure and women’s welfare and well-being, and reductions in social barriers and prejudice facing women’s participation in the labour market (Mammen and Paxson 2000; Cuberes and Teignier 2014; Tam 2011; Verme 2015). In this context, FDI would further exercise indirect positive influences on the female employment rate in the economy and promote the transitional effects on the share of female employment across sectors through its role as a catalyst for economic growth and development. The transition of the female labour force across sectors could be partially observed in employment statistics, accompanying higher returns to education and higher school enrolment rates of women in an open economy (Wacker et al. 2017).

Inward FDI could induce reductions in the female employment rate over some stages of economic development. These impacts could be outlined through the relationship between economic growth and female labour force participation (Goldin 1995; Luci 2009; Tam 2011). The income effect generates decreases in female labour force participation rates, stemming from rising demand for children in the early stages of economic development, where agriculture plays a significant role in the economy. Also, the introduction of machinery would give a comparative advantage to male workers, leading to declines in the demand (and employment) of female workers through the substitution effect. However, with further economic growth and expansion of the service sector, cognitive abilities become more prevalent in economic activities, and the demand for female workers increases through more-cognitive/less-physical employment positions. Also, in labour-abundant countries, economic growth is generally marked by the expansion of the labour-intensive manufacturing sector, accompanied by increasing demand for unskilled female workers.

Foreign affiliates of MNEs often have higher propensities to employ female workers compared to domestic firms in host countries as they tend to benefit from under-utilized female labour resources, a compliant female labour force, gender wage gap, and absence of (or inefficient) gender equality regulations (Ozler 2000; Oostendorp 2009; Standing 2010).4 This channel is particularly prevalent in the case of export-oriented foreign affiliates that seek to employ low-wage, unskilled female workers in developing countries where gender wage gaps prevail. Many empirical studies (e.g., Fernández-Kelly 1983; Chen et al. 2013; Coniglio et al. 2017; Fernandes and Kee 2020) show that foreign affiliates of MNEs tend to disproportionally employ unskilled female workers in developing countries to benefit from the gender wage gap and compliant female labour force.5

The role of FDI in spreading new cultural norms and business practices in the host country, including those associated with women in the workplace, is well established in the literature (Watson 2006; Lawler and Bae 1998; Monge-González et al. 2021). In this context, foreign affiliates of MNEs often adopt corporate social responsibility schemes and gender-inclusive policies that tend to reduce gender inequality in the workplace (Kucera 2002; Kodama et al. 2018). Lawler and Bae (1998) underline the role of MNEs in reducing gender-based employment discrimination in developing countries. They attribute this role to the cultural factors that are expressed through the existing wedge in gender equality levels between the source—developed countries—and the host—developing countries. Several studies (e.g., United Nations Conference on Trade and Development (UNCTAD) 2014; Tang and Zhang 2021; Choi and Greaney 2022) present empirical evidence that emphasizes the cultural and socio-economic characteristics of FDI’s source country in determining the effects on female employment in the host country. These studies find that foreign affiliates of MNEs originating from countries with higher levels of gender equality feature higher female employment rates and narrower gender wage gaps, and they tend to generate cultural spillovers to domestic firms.6

MNEs are generally more resistant to the implications of gender-biased social norms when undertaking FDI in host countries. This relative immunity is transmitted into the business culture of their foreign affiliates, eventually spilling over into the hiring practices of domestic firms in host countries (Lawler and Bae 1998; Monge-González et al. 2021). Also, managers of foreign affiliates of MNEs are often expatriated from MNEs’ home countries or selected among nationals that tend to have an affinity to the culture of MNEs’ home countries (e.g., nationals that completed education in developed countries). As such, they tend to be more resistant to the existing discriminatory cultural norms and practices that prevail in host countries (Lawler and Bae 1998). Monge-González et al. (2021) describe the cultural transmission mechanisms from foreign affiliates of MNEs to domestic firms by (1) the demonstration effect through which social norms in foreign affiliates are emulated and infused into the work environment of domestic firms, and (2) the learning effect that mainly occurs through the mobility of workers from foreign affiliates to domestic firms or through other business networks.

Foreign affiliates of MNEs do not normally bend to the social norms of the host country since such inclinations would be profit-decreasing and inefficient. This point is emphasized by Siegel et al. (2019), who indicate that the neoclassical theories where foreign affiliates abide by social norms that exist in the host country do not account for the advantages of non-conformism to those norms, which allow foreign affiliates to benefit from the talents and work contributions of the female labour force.

Becker (1957) hypothesizes that increases in market competition tend to lessen the extent of employment discrimination since such practices would be particularly costly to the employing firms. Hence, these firms would be compelled to overcome discriminatory cultural norms and organizational practices to survive market competition (Chen et al. 2013; Heyman et al. 2013; Vahter and Masso 2019). Foreign affiliates of MNEs are usually tightly connected to the competitive global market, rendering gender-biased discriminatory practices intrinsically detrimental to their performance. Furthermore, increases in FDI inflows typically lead to rises in market competition in the host country, inducing domestic firms to lessen their gender-biased practices and promoting better allocation of labour resources (Black and Brainerd 2004; Chen et al. 2013). Hence, this supplementary channel would generate positive effects of inward FDI on gender equality in the workplace and on female employment.

Some studies (e.g., Aguayo-Tellez 2012; Juhn et al. 2014; Vahter and Masso 2019) underline that FDI constitutes a significant vector that transmits skill-intensive technology through the foreign affiliates of MNEs. This channel tends to be marked by higher complementarity levels with female workers since women feature, in general, comparative advantage in cognitive skills vis-à-vis physical skills. The demand for the cognitive abilities of workers would rise with technology transfer, whereas the demand for the physical abilities of workers, which are often expressed through tasks performed by men, tends to decrease. Also, upgraded technologies that are introduced by foreign affiliates into the host country could horizontally and vertically spill over into other domestic firms (Blomström et al. 2003; United Nations Conference on Trade and Development (UNCTAD) 2014; Fernandes and Kee 2020). In such cases, the complementarity between new technologies and female employment would be further emphasized in the economy.

It is worth noting that FDI in one sector could have indirect effects on female employment in domestic firms with upward or downward linkages to foreign affiliates of MNEs through the supply chain (Saadi 2010; Fernandes and Kee 2020). In many cases, inward FDI leads to significant growth of the industrial sector in host countries. This growth is normally accompanied by a rise in demand for sub-contracted services offered by some specific service sectors (e.g., information technology, financial services) or administrative positions within domestic firms. The more-cognitive and less-physical nature of such services aligns with women’s comparative advantage, eventually leading to increases in female employment rates in these sectors.

There are, however, some counteracting microeconomic factors through which inward FDI decreases female employment rates. In some cases, FDI brings capital-intensive production processes and advanced technologies that tend to replace unskilled labour in host countries. Female workers, who often make up larger proportions of the unskilled labour force in some labour-intensive sectors (e.g., export-oriented manufacturing industries in developing countries), will be significantly affected and, as such, female employment will be disproportionally displaced (Tejani and Milberg 2016; Pham et al. 2021). Also, foreign affiliates of MNEs typically require more skilled labour compared to domestic firms to match their technological endowment. Then, their production conditions could imply lower demand for unskilled female workers, particularly in the industrial sector (Tejani and Milberg 2016).

3. Empirical Model

The empirical analysis aims to estimate the effects of inward FDI on the national female employment rate and to determine its role in the transition of female labour across sectors (i.e., the agricultural, industrial, and service sectors). The primary hypothesis posits that inward FDI promotes the transition of women from agriculture through the industrial sector toward the service sector. As such, a negative relationship is expected between inward FDI and the share of female employment in the agriculture sector, whereas a positive relationship is expected between inward FDI and the share of female employment in the service sector. Meanwhile, the FDI effect on the share of female employment in the industrial sector is indeterminate, contingent upon various market and economic conditions, along with the relative inflow from the agricultural sector and outflow toward the service sector. The dynamic panel specifications are estimated through the GMM System estimator that accounts for country-specific unobserved heterogeneity and that involves the estimation of a system of first-difference and level equations (Arellano and Bover 1995; Blundell and Bond 1998).7 Then, lags in the level form are used as instruments in the first-difference equations, whereas first differences are used as instruments in the level equations. Letting the superscript “$e$” denote the overall national economy, ${FEmp}_{it}^e$ represents the national female employment rate in country $i$ at time $t$. The corresponding dynamic panel equation is given by the following:

$${FEmp}_{it}^e={\theta }^e+{\beta }^e{FEmp}_{it-1}^e+{\gamma }^e{FDIOp}_{it}+{{\mathit{\delta }}^{\mathit{e}}}^{\prime }{\mathit{X}}_{\mathit{i}\mathit{t}}+{\omega }_t^e+{\eta }_i^e+{\vartheta }_{it}^e$$

(1)

where ${FDIOp}_{it}$ depicts FDI openness, and ${\mathit{X}}_{\mathit{i}\mathit{t}}$ is a vector of regressors comprising standard control variables (e.g., fertility rate, population size, economic growth rate, school enrolment rate, inflation rate). The empirical model includes the time-specific effect ${\omega }_t^e$ that controls for time-specific shocks. Also, ${\eta }_i^e$ is the unobserved country-specific effect, and ${\vartheta }_{it}^e$ is the remaining stochastic term. The GMM System estimation involves the following set of moment conditions for the first-difference and level equations:

$$\left\{\begin{array}{c}E\left[{FEmp}_{it-z}^e·{∆\vartheta }_{it}^e\right]=0 \mathrm{for} z\ge 2\\ E\left[{FDIOp}_{it-z}·{∆\vartheta }_{it}^e\right]=0 \mathrm{for} z\ge 1\\ E\left[x_{it-z}·{∆\vartheta }_{it}^e\right]=0 \mathrm{for} x_{it}\in {\mathit{X}}_{\mathit{i}\mathit{t}} \mathrm{and} z\ge 1\end{array}\right.$$

(2)

$$\left\{\begin{array}{c}E\left[{∆FEmp}_{it-z}^e·\left({\eta }_i^e+{\vartheta }_{it}^e\right)\right]=0 \mathrm{for} z\ge 2\\ E\left[∆{FDIOp}_{it-z}·\left({\eta }_i^e+{\vartheta }_{it}^e\right)\right]=0 \mathrm{for} z\ge 1\\ E\left[{∆x}_{it-z}·\left({\eta }_i^e+{\vartheta }_{it}^e\right)\right]=0 \mathrm{for} x_{it}\in {\mathit{X}}_{\mathit{i}\mathit{t}} \mathrm{and} z\ge 1\end{array}\right.$$

(3)

Next, let $k$ denote a given sector with $k=a$ for the agricultural sector, $k=m$ for the industrial sector, and $k=s$ for the service sector. Let $Sh{FEmp}_{it}^k$ represent the share of female employment in sector $k$ in country $i$ at time $t$, with ${\sum }_{k}Sh{FEmp}_{it}^k=1$ $\forall i,t$. The basic dynamic empirical model is determined as follows:

$$\left\{\begin{array}{c}{ShFEmp}_{it}^a={\theta }^a+{\beta }^a{ShFEmp}_{it-1}^a+{\gamma }^a{FDIOp}_{it}+{{\mathit{\delta }}^{\mathit{a}}}^{\prime }{\mathit{X}}_{\mathit{i}\mathit{t}}+{\omega }_t^a+{\eta }_i^a+{\vartheta }_{it}^a \\ {ShFEmp}_{it}^m={\theta }^m+{\beta }^m{ShFEmp}_{it-1}^m+{\gamma }^m{FDIOp}_{it}+{{\mathit{\delta }}^{\mathit{m}}}^{\prime }{\mathit{X}}_{\mathit{i}\mathit{t}}+{\omega }_t^m+{\eta }_i^m+{\vartheta }_{it}^m\\ {ShFEmp}_{it}^s={\theta }^s+{\beta }^s{ShFEmp}_{it-1}^s+{\gamma }^s{FDIOp}_{it}+{{\mathit{\delta }}^{\mathit{s}}}^{\prime }{\mathit{X}}_{\mathit{i}\mathit{t}}+{\omega }_t^s+{\eta }_i^s+{\vartheta }_{it}^s \end{array}\right.$$

(4)

where ${\omega }_t^k$ is the time-specific effect, and ${\eta }_i^k$ and ${\vartheta }_{it}^k$ represent the unobserved country-specific effect and the remaining stochastic term, respectively. The corresponding sets of moment conditions are as follows:

$$\left\{\begin{array}{c}E\left[{ShFEmp}_{it-z}^k·{∆\vartheta }_{it}^k\right]=0 \mathrm{for} z\ge 2\\ E\left[{FDIOp}_{it-z}·{∆\vartheta }_{it}^k\right]=0 \mathrm{for} z\ge 1\\ E\left[x_{it-z}·{∆\vartheta }_{it}^k\right]=0 \mathrm{for} x_{it}\in {\mathit{X}}_{\mathit{i}\mathit{t}} \mathrm{and} z\ge 1\end{array}\right.$$

(5)

$$\left\{\begin{array}{c}E\left[{∆ShFEmp}_{it-z}^k·\left({\eta }_i^k+{\vartheta }_{it}^k\right)\right]=0 \mathrm{for} z\ge 2\\ E\left[∆{FDIOp}_{it-z}·\left({\eta }_i^k+{\vartheta }_{it}^k\right)\right]=0 \mathrm{for} z\ge 1\\ E\left[{∆x}_{it-z}·\left({\eta }_i^k+{\vartheta }_{it}^k\right)\right]=0 \mathrm{for} x_{it}\in {\mathit{X}}_{\mathit{i}\mathit{t}} \mathrm{and} z\ge 1\end{array}\right.$$

(6)

The estimated coefficients on the FDI openness variable depict the short-run effects, whereas the corresponding long-run (enduring) effects are determined by ${\gamma }^e/\left(1-{\beta }^e\right)$ in the case of the national female employment rate (${FEmp}_{it}^e$) and by ${\gamma }^k/\left(1-{\beta }^k\right)$ in the case of the share of female employment in sector $k$ ($Sh{FEmp}_{it}^k$). The long-run effects account for the cumulative effects of FDI openness on the female employment variables. Notably, there is a high persistence over time when ${\gamma }^e$ and ${\gamma }^k$ are closer to one, over the process of moving toward the long-run equilibrium. The GMM System estimations are assessed by the following two standard tests (Arellano and Bond 1991; Arellano and Bover 1995; Blundell and Bond 1998): the Sargan–Hansen (SH) test of over-identifying restrictions for the joint validity of instruments, and the Arellano–Bond test of second-order autocorrelation AR(2) for the consistency of the estimates.

The empirical analysis is implemented using a panel dataset that spans the time period 2001–2019 and covers 148 countries (for a total of 2812 observations).8 The datasets on the female employment rate in the economy and the share of female employment across sectors are derived from the International Labour Organization’s database (ILOSTAT), where employment is defined as “persons of working age who, during a short reference period, were engaged in any activity to produce goods or provide services for pay or profit, whether at work during the reference period (i.e., who worked in a job for at least one hour) or not at work due to temporary absence from a job, or to working-time arrangements”. The working-age population in this database is defined as those aged 15 and older. The variable ${FEmp}_{it}^e$ represents the women’s employment-to-population ratio (expressed in percentage terms), designating the proportion of a country’s female population that is employed. The variables ${ShFEmp}_{it}^a$, ${ShFEmp}_{it}^m$, and ${ShFEmp}_{it}^s$ represent the percentage share of female employment in the agricultural, industrial, and service sectors, respectively, in total female employment.9 

Table 1. Descriptive statistics.

	(i)	(ii)	(iii)	(iv)
	Mean	St. Dev.	Min.	Max.
${FEmp}_{it}^e$	48.15	15.77	4.49	86.01
${ShFEmp}_{it}^a$	27.24	27.47	0.01	96.50
${ShFEmp}_{it}^m$	12.08	6.88	0.33	37.52
${ShFEmp}_{it}^s$	60.68	25.59	2.79	98.67
${FDIOp}_{it}$	5.42	17.97	−40.29	449.08
${FDI}_{it}$	11.10	39.75	−296.27	758.61
${ForOwnOp}_{it}$	5.55	3.77	0	10
${CapitalFr}_{it}$	3.66	2.92	0	10
${ForVisitFr}_{it}$	5.11	3.50	0	10

Notes: The descriptive statistics in this table cover a total of 2812 observations and include the following variables: ${FEmp}_{it}^e$ (female employment-to-population ratio, %); ${ShFEmp}_{it}^a$ (share of female employment in the agricultural sector, % of female employment); ${ShFEmp}_{it}^m$ (share of female employment in the industrial sector, % of female employment); ${ShFEmp}_{it}^s$ (share of female employment in the service sector, % of female employment); ${FDIOp}_{it}$ (FDI inflows to GDP ratio, %); ${FDI}_{it}$ (FDI inflows in constant USD 2010, billion); ${ForOwnOp}_{it}$ (foreign ownership/investment openness, indicator); ${CapitalFr}_{it}$ (capital freedom, indicator); and ${ForVisitFr}_{it}$ (freedom of foreigners to visit, indicator).

presents descriptive statistics for the overall female employment rate in the economy and the share of female employment across sectors.

The basic FDI openness measure (${FDIOp}_{it}$) is represented through the ratio of net inflows of FDI to GDP, where FDI refers to direct investment equity flows in the reporting economy and covers the sum of equity capital, reinvestment of earnings, and other capital.10 This FDI dataset is derived from the Balance of Payment (BoP) database of the International Monetary Fund (IMF). The benchmark empirical analysis also uses alternative indicators measuring the extent of foreign capital openness since sector-specific inward FDI datasets are scarce and only available for a subset of (mostly developed) countries. These indicators are sourced from the Fraser Institute’s database, and they comprise the foreign ownership/investment openness index (${ForOwnOp}_{it}$), capital freedom index (${CapitalFr}_{it}$), and freedom of foreigners to visit index (${ForVisitFr}_{it}$). These indicators range between zero and 10, where higher values indicate lower restrictions (or higher freedom levels) (Gwartney et al. 2021).11 Table 1 displays descriptive statistics for the benchmark FDI openness measure, total FDI inflows (in constant USD 2010), and FDI freedom indicators. The empirical specification includes conventional control variables in the female employment equations, covering macroeconomic factors [economic growth rate (${EconGrowth}_{it}$), unemployment rate (${Unemp}_{it}$), and inflation rate (${Inflation}_{it}$)] and socio-economic/socio-demographic factors [fertility rate (${Fertility}_{it}$), log of population size (${lnPopulation}_{it}$), and primary school enrolment rate (${PrSEnrol}_{it}$)]. The corresponding datasets are derived from the World Bank database.

4. Benchmark Empirical Results

The empirical results from the benchmark dynamic panel regressions are presented in

Table 2. Effect of FDI on female employment rate.

	(i)	(ii)	(iii)	(iv)	(v)
	GMM System	GMM System [FOD]	GMM System [Two-Step]	GMM System	GMM System [Two-Step]
${FEmp}_{it-1}^e$	0.855 a	0.849 a	0.860 a	0.847 a	0.842 a
	(0.023)	(0.026)	(0.024)	(0.024)	(0.025)
${FDIOp}_{it}$	0.092	0.105	0.087	0.064	0.068
	(0.070)	(0.076)	(0.069)	(0.073)	(0.076)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				0.070 b	0.066 b
				(0.028)	(0.029)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				0.045	0.042
				(0.030)	(0.032)
${Fertility}_{it}$	−0.096	−0.089	−0.093	−0.086	−0.083
	(0.075)	(0.077)	(0.076)	(0.076)	(0.078)
${lnPopulation}_{it}$	−0.081	−0.078	−0.065	−0.092	−0.084
	(0.065)	(0.070)	(0.068)	(0.069)	(0.074)
${EconGrowth}_{it}$	1.710 b	1.654 b	1.904 b	2.049 b	2.103 b
	(0.817)	(0.852)	(0.875)	(0.895)	(0.870)
${PrSEnrol}_{it}$	0.017	0.018	0.017	0.020	0.022
	(0.012)	(0.013)	(0.012)	(0.013)	(0.014)
${Inflation}_{it}$	−0.005	−0.007	−0.004	−0.003	−0.004
	(0.006)	(0.007)	(0.006)	(0.006)	(0.007)
${Unemp}_{it}$	−0.068 b	−0.079 b	−0.070 b	−0.056 c	−0.070 c
	(0.031)	(0.035)	(0.033)	(0.033)	(0.036)
	Long-Run Effects
${FDIOp}_{it}$	0.634	0.695	0.621	0.418	0.430
	(0.479)	(0.498)	(0.483)	(0.474)	(0.477)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				0.458 b	0.418 b
				(0.182)	(0.181)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				0.294	0.266
				(0.196)	(0.202)
$N\times T$	2812	2812	2812	2812	2812
$\mathrm{S}\mathrm{H} \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>\chi 2)$	0.451	0.455	0.449	0.487	0.481
$\mathrm{A}\mathrm{R}\left(2\right) \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>z)$	0.628	0.639	0.625	0.596	0.589

Notes: The dependent variable is ${FEmp}_{it}^e$ (female employment-to-population ratio, %). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively. The $\mathrm{S}\mathrm{H}$ test is the Sargan–Hansen test of instrument over-identification restrictions. The $\mathrm{A}\mathrm{R}\left(2\right)$ test is the Arellano–Bond test for second-order serial correlation. Columns (iv) and (v) present the results when specifying the effect of ${FDIOp}_{it}$ by the RGDPC tercile categories. The third tercile (${Trc}^3$) is set as the reference, whereas the first and second terciles (${Trc}^1$ and ${Trc}^2$) are presented as deviations from the reference.

for the women’s employment-to-population ratio (${FEmp}_{it}^e$), and in

Table 3. Effect of FDI on the share of female employment in the agricultural sector.

	(i)	(ii)	(iii)	(iv)	(v)
	GMM System	GMM System [FOD]	GMM System [Two-Step]	GMM System	GMM System [Two-Step]
${ShFEmp}_{it-1}^a$	0.838 a	0.830 a	0.842 a	0.833 a	0.836 a
	(0.026)	(0.028)	(0.025)	(0.028)	(0.027)
${FDIOp}_{it}$	−0.349 a	−0.352 a	−0.335 a	−0.247 a	−0.239 a
	(0.082)	(0.084)	(0.081)	(0.083)	(0.078)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				−0.084 a	−0.077 a
				(0.025)	(0.026)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				−0.126 a	−0.118 a
				(0.022)	(0.024)
${Fertility}_{it}$	0.741 a	0.767 a	0.746 a	0.705 a	0.695 a
	(0.214)	(0.225)	(0.217)	(0.219)	(0.225)
${lnPopulation}_{it}$	0.191 c	0.198 c	0.185 c	−0.229 b	−0.233 b
	(0.100)	(0.106)	(0.104)	(0.111)	(0.117)
${EconGrowth}_{it}$	−2.592 a	−2.711 a	−2.455 a	−2.393 a	−2.272 a
	(0.885)	(0.954)	(0.891)	(0.855)	(0.864)
${PrSEnrol}_{it}$	−0.018	−0.009	−0.020	−0.020	−0.016
	(0.014)	(0.014)	(0.014)	(0.015)	(0.014)
${Inflation}_{it}$	0.007	0.009	0.006	0.013	0.009
	(0.008)	(0.009)	(0.008)	(0.009)	(0.008)
${Unemp}_{it}$	−0.050 c	−0.056 c	−0.042	−0.039	−0.030
	(0.027)	(0.029)	(0.026)	(0.028)	(0.026)
	Long-Run Effects
${FDIOp}_{it}$	−2.154 a	−2.071 a	−2.120 a	−1.479 a	−1.457 a
	(0.559)	(0.548)	(0.563)	(0.539)	(0.521)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				−0.503 a	−0.470 a
				(0.162)	(0.174)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				−0.754 a	−0.720 a
				(0.142)	(0.159)
$N\times T$	2812	2812	2812	2812	2812
$\mathrm{S}\mathrm{H} \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>\chi 2)$	0.429	0.411	0.425	0.505	0.516
$\mathrm{A}\mathrm{R}\left(2\right) \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>z)$	0.394	0.401	0.399	0.432	0.438

Notes: The dependent variable is ${ShFEmp}_{it}^a$ (share of female employment in the agricultural sector, % of female employment). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively. The $\mathrm{S}\mathrm{H}$ test is the Sargan–Hansen test of instrument over-identification restrictions. The $\mathrm{A}\mathrm{R}\left(2\right)$ test is the Arellano–Bond test for second-order serial correlation. Columns (iv) and (v) present the results when specifying the effect of ${FDIOp}_{it}$ by the RGDPC tercile categories. The third tercile (${Trc}^3$) is set as the reference, whereas the first and second terciles (${Trc}^1$ and ${Trc}^2$) are presented as deviations from the reference.

,

Table 4. Effect of FDI on the share of female employment in the industrial sector.

	(i)	(ii)	(iii)	(iv)	(v)
	GMM System	GMM System [FOD]	GMM System [Two-Step]	GMM System	GMM System [Two-Step]
${ShFEmp}_{it-1}^m$	0.892 a	0.884 a	0.889 a	0.865 a	0.871 a
	(0.024)	(0.025)	(0.024)	(0.026)	(0.027)
${FDIOp}_{it}$	−0.027	−0.020	−0.025	−0.011	−0.014
	(0.041)	(0.040)	(0.038)	(0.043)	(0.044)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				−0.007	−0.010
				(0.015)	(0.016)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				−0.034 b	−0.030 c
				(0.016)	(0.016)
${Fertility}_{it}$	0.120 c	0.115 c	0.118 c	0.128 c	0.130 c
	(0.064)	(0.065)	(0.062)	(0.072)	(0.075)
${lnPopulation}_{it}$	0.168 b	0.165 b	−0.155 b	0.200 b	0.194 b
	(0.075)	(0.075)	(0.074)	(0.081)	(0.085)
${EconGrowth}_{it}$	2.845 a	2.902 a	2.970 a	2.409 a	2.546 a
	(0.789)	(0.784)	(0.841)	(0.782)	(0.835)
${PrSEnrol}_{it}$	0.009	0.007	0.009	0.011	0.011
	(0.007)	(0.007)	(0.007)	(0.008)	(0.008)
${Inflation}_{it}$	0.005	0.006	0.007	0.002	0.001
	(0.005)	(0.005)	(0.005)	(0.006)	(0.006)
${Unemp}_{it}$	−0.009	−0.010	−0.010	−0.012	−0.012
	(0.013)	(0.009)	(0.013)	(0.013)	(0.014)
	Long-Run Effects
${FDIOp}_{it}$	−0.250	−0.172	−0.225	−0.081	−0.108
	(0.379)	(0.342)	(0.341)	(0.318)	(0.338)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				−0.052	−0.077
				(0.113)	(0.123)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				−0.252 b	−0.231 c
				(0.118)	(0.121)
$N\times T$	2812	2812	2812	2812	2812
$\mathrm{S}\mathrm{H} \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>\chi 2)$	0.490	0.474	0.494	0.550	0.556
$\mathrm{A}\mathrm{R}\left(2\right) \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>z)$	0.683	0.672	0.680	0.658	0.649

Notes: The dependent variable is ${ShFEmp}_{it}^m$ (share of female employment in the industrial sector, % of female employment). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively. The $\mathrm{S}\mathrm{H}$ test is the Sargan–Hansen test of instrument over-identification restrictions. The $\mathrm{A}\mathrm{R}\left(2\right)$ test is the Arellano–Bond test for second-order serial correlation. Columns (iv) and (v) present the results when specifying the effect of ${FDIOp}_{it}$ by the RGDPC tercile categories. The third tercile (${Trc}^3$) is set as the reference, whereas the first and second terciles (${Trc}^1$ and ${Trc}^2$) are presented as deviations from the reference.

and

Table 5. Effect of FDI on the share of female employment in the service sector.

	(i)	(ii)	(iii)	(iv)	(v)
	GMM System	GMM System [FOD]	GMM System [Two-Step]	GMM System	GMM System [Two-Step]
${ShFEmp}_{it-1}^s$	0.853 a	0.850 a	0.856 a	0.848 a	0.851 a
	(0.029)	(0.030)	(0.030)	(0.032)	(0.034)
${FDIOp}_{it}$	0.437 a	0.442 a	0.414 a	0.315 a	0.294 a
	(0.075)	(0.077)	(0.075)	(0.078)	(0.072)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				0.070 a	0.059 a
				(0.022)	(0.022)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				0.112 a	0.104 a
				(0.024)	(0.027)
${Fertility}_{it}$	−0.577 a	−0.613 a	0.598 a	−0.539 a	−0.520 a
	(0.178)	(0.187)	(0.181)	(0.184)	(0.187)
${lnPopulation}_{it}$	−0.366 b	−0.378 b	−0.360 b	−0.381 b	−0.354 b
	(0.154)	(0.161)	(0.153)	(0.159)	(0.163)
${EconGrowth}_{it}$	1.352 c	1.425 c	1.198 c	1.560 b	1.515 b
	(0.712)	(0.716)	(0.717)	(0.734)	(0.742)
${PrSEnrol}_{it}$	0.010	0.009	0.011	0.011	0.009
	(0.013)	(0.013)	(0.013)	(0.013)	(0.014)
${Inflation}_{it}$	−0.007	−0.008	−0.009	−0.010	−0.012
	(0.010)	(0.011)	(0.010)	(0.010)	(0.010)
${Unemp}_{it}$	0.040	0.047	0.041	0.035	0.040
	(0.029)	(0.032)	(0.030)	(0.031)	(0.032)
	Long-Run Effects
${FDIOp}_{it}$	2.973 a	2.947 a	2.875 a	2.072 a	1.973 a
	(0.553)	(0.556)	(0.573)	(0.563)	(0.532)
${d\_FDIOp}_{it} \left({Trc}^1\right)$				0.461 a	0.396 a
				(0.158)	(0.153)
${d\_FDIOp}_{it} \left({Trc}^2\right)$				0.737 a	0.698 a
				(0.180)	(0.202)
$N\times T$	2812	2812	2812	2812	2812
$\mathrm{S}\mathrm{H} \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>\chi 2)$	0.387	0.390	0.396	0.435	0.454
$\mathrm{A}\mathrm{R}\left(2\right) \mathrm{T}\mathrm{e}\mathrm{s}\mathrm{t} (Pr>z)$	0.313	0.323	0.310	0.295	0.283

Notes: The dependent variable is ${ShFEmp}_{it}^s$ (share of female employment in the service sector, % of female employment). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively. The $\mathrm{S}\mathrm{H}$ test is the Sargan–Hansen test of instrument over-identification restrictions. The $\mathrm{A}\mathrm{R}\left(2\right)$ test is the Arellano–Bond test for second-order serial correlation. Columns (iv) and (v) present the results when specifying the effect of ${FDIOp}_{it}$ by the RGDPC tercile categories. The third tercile (${Trc}^3$) is set as the reference, whereas the first and second terciles (${Trc}^1$ and ${Trc}^2$) are presented as deviations from the reference.

for the share of female employment in the agricultural, industrial, and service sectors (${ShFEmp}_{it}^a$, ${ShFEmp}_{it}^m$, and ${ShFEmp}_{it}^s$), respectively. The SH tests for over-identifying restrictions and the AR(2) tests for the prevalence of second-order serial correlation do not reject the corresponding null hypotheses, and they confirm the validity of the estimates across the empirical models in these tables. Columns (i) in these tables show the results from the one-step GMM System estimations. The effect of FDI openness (${FDIOp}_{it}$) on the female employment rate (${FEmp}_{it}^e$) is not statistically significant. This outcome may not necessarily imply that inward FDI does not have any impact on female employment, but it could rather depict the net effects of counteracting macroeconomic and microeconomic implications within and across sectors.

A closer look into columns (i) of the following tables reveals that the effect of FDI openness on the share of female employment in the agricultural sector (${ShFEmp}_{it}^a$) is negative and statistically significant at the 1% level. The estimates in Table 3 indicate that a two-fold increase in FDI openness would lead to a decrease in the share of female employment in the agricultural sector by 0.35 percentage points (pps) in the short run and by [0.349/(1 − 0.838)] = 2.15 pps in the long run, ceteris paribus. In contrast, the effects of FDI openness on the share of female employment in the service sector (${ShFEmp}_{it}^s$) are positive and statistically significant at the 1% level. The estimates in Table 5 indicate that a two-fold rise in FDI openness leads to an increase in the share of female employment in the service sector by 0.44 pps in the short run and by [0.437/(1 − 0.853)] = 2.97 pps in the long run, ceteris paribus. Table 4 shows that the effect of FDI openness on the share of female employment in the industrial sector (${ShFEmp}_{it}^m$) is statistically insignificant. These outcomes could stem from the net effects of counteracting factors as female employment shifts from agriculture through the industrial sector toward the service sector.

Columns (ii) of these tables present the results when carrying out the GMM System estimation with Forward Orthogonal Deviations (FODs) instead of first-differencing. The FOD approach consists of subtracting the average of all available future observations rather than subtracting the previous observations (Arellano and Bover 1995). Also, columns (iii) of these tables present the results from the two-step GMM System estimations, where the covariance matrix is subjected to finite-sample correction to tackle downward bias in standard errors (Windmeijer 2005). The estimates are found to be comparable to the benchmark estimates in columns (i) across these tables.

The promoting effects of inward FDI on economic development and restructuring (Borensztein et al. 1998; Alfaro et al. 2004) would foster the transition of women from agriculture toward the industrial and service sectors (Boserup 1970; Goldin 1995). While the net effect on the overall female employment rate in the economy is not statistically significant, the findings are consistent with the prevalence of macroeconomic channels through which inward FDI impacts the share of female employment across sectors. The microeconomic factors are also encompassed within the estimated effects. For instance, foreign affiliates tend to benefit from the gender wage gap, leading to higher female employment rates (Oostendorp 2009). Also, they often characterize cultural and business norms that encourage gender equality in the workplace (Lawler and Bae 1998) and bring about increases in market competition that reduce gender-based discriminatory practices (Chen et al. 2013). However, there could be some counteracting implications when capital-intensive production processes and advanced technologies introduced by foreign affiliates of MNEs accompanied by higher demand for skilled labour disproportionally replace unskilled female workers. The results complement the empirical literature that underlines the relationships between female employment and inward FDI (and the corresponding activities of MNEs) (e.g., Chen et al. 2013; Kodama et al. 2018; Siegel et al. 2019), and they are consistent with the growth-led transition of female employment across sectors (Boserup 1970; Fuchs 1980; Goldin 1995; Foster and Rosenzweig 2007; Ghazalian 2022).

The significant positive effects of inward FDI on the share of female employment in the service sector could naturally follow the outcomes from stimulated economic development that enhances women’s socio-economic status, and that transitions them from low-skill manufacturing employment toward more cognitive, administrative, and managerial occupations. They could also be associated with divergence over the microeconomic conduits across sectors. For instance, cultural and business norms of gender equality could be more prevalent and practiced across foreign affiliates in the service sector compared to those in the manufacturing sector. The pronounced adverse impacts of FDI openness on the share of female employment in agriculture underscore the imperative for economic and social development policies aimed at enhancing economic opportunities for women in agriculture, thereby improving their well-being and social status to mitigate outward migration and preserve women’s economic and social roles in rural areas.

The empirical analysis proceeds to examine whether the effect of FDI openness on female employment exhibits variation with national income level, which is proxied by the Real GDP per Capita (RGDPC) of the host country. Hence, the coefficients on ${FDIOp}_{it}$ are estimated next by the RGDPC tercile categories: ${Trc}^1$ (the lowest tercile), ${Trc}^2$ (the middle tercile), and ${Trc}^3$ (the highest tercile). The latter is set as the reference group through the regressions. The analysis is carried out by contrasting the effects among tercile categories, and the estimates are presented as deviations from the effect of FDI openness on female employment of the reference group. The corresponding results are presented in columns (iv) and (v) for the one-step and two-step GMM System estimations, respectively (see Appendix A for the estimates of the total effects for ${Trc}^1$ and ${Trc}^2$).12 They are found to be comparable and, therefore, the discussion follows the one-step estimates.

The results show that the effects of FDI openness on women’s employment-to-population ratio (${FEmp}_{it}^e$) are positive but relatively small in the case of ${Trc}^1$, and statistically insignificant in the cases of the other categories. The empirical analysis across sectors reveals some variations. In the case of the agricultural sector, the negative short-run effects of FDI openness on the share of female employment are moderately larger (in absolute terms) in the cases of ${Trc}^1$ and ${Trc}^2$ standing at −0.33 pps and −0.37 pps, respectively, compared to −0.25 pps in the case of ${Trc}^3$. The corresponding long-run estimates are −1.98 pps, −2.23 pps, and −1.48 pps, respectively. Meanwhile, in the case of the service sector, the estimates indicate that the positive effects of FDI openness on the share of female employment are moderately larger in the cases of the low and middle tercile categories. Specifically, they indicate that a two-fold increase in FDI openness would lead to short-run increases in the share of female employment in the service sector by 0.385 pps and 0.427 pps in the cases of ${Trc}^1$ and ${Trc}^2$, respectively, compared to 0.315 pps in the case of ${Trc}^3$. The corresponding long-run estimates stand at increases by 2.53 pps, 2.81 pps, and 2.07 pps, respectively. The estimates for the industrial sector show small statistically significant effects of FDI openness in the case of ${Trc}^2$. Overall, these results suggest that, starting from low national income levels, the effects of FDI openness on female employment moderately increase with economic growth, and that these effects slow down at higher national income levels as the propensity of transition across sectors is broadly realized.

5. Robustness Checks

The empirical analysis proceeds to substitute the original FDI openness variable with alternative FDI openness indicators in the empirical specifications. As noted earlier, these indicators are collected from the Fraser Institute’s database, and they comprise an overall FDI openness index ($x{FDIOp}_{it}$), and sub-component indicators covering foreign ownership/investment openness (${ForOwnOp}_{it}$), capital freedom (${CapitalFr}_{it}$), and freedom of foreigners to visit (${ForVisitFr}_{it}$). The one-step and two-step GMM System estimates of these indicators for the female employment-to-population rate and female employment share across sectors are presented in

Table 6. Empirical results—FDI openness indicators.

	(i)	(ii)	(iii)	(iv)
	GMM System		GMM System [Two-Step]
	Short-Run Effect	Long-Run Effect	Short-Run Effect	Long-Run Effect
	$\mathrm{Female} \mathrm{Employment} \mathrm{Rate} ({FEmp}_{it}^e$)
${ForOwnOp}_{it}$	0.096 c	0.744 c	0.092 c	0.736 c
	(0.046)	(0.350)	(0.049)	(0.392)
${CapitalFr}_{it}$	0.043	0.319	0.041	0.297
	(0.027)	(0.200)	(0.028)	(0.203)
${ForVisitFr}_{it}$	0.026	0.188	0.025	0.180
	(0.025)	(0.180)	(0.024)	(0.172)
$x{FDIOp}_{it}$	0.064	0.457	0.060	0.423
	(0.040)	(0.283)	(0.043)	(0.302)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Agricultural} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^a$)
${ForOwnOp}_{it}$	−0.408 a	−2.566 a	−0.395 a	−2.582 a
	(0.095)	(0.654)	(0.090)	(0.636)
${CapitalFr}_{it}$	−0.240 a	−1.437 a	−0.233 a	−1.421 a
	(0.059)	(0.382)	(0.057)	(0.374)
${ForVisitFr}_{it}$	−0.095 b	−0.562 b	−0.090 b	−0.529 b
	(0.044)	(0.266)	(0.042)	(0.251)
$x{FDIOp}_{it}$	−0.313 a	−1.841 a	−0.308 a	−1.801 a
	(0.078)	(0.479)	(0.074)	(0.463)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Industrial} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^m$)
${ForOwnOp}_{it}$	−0.043 c	−0.387 c	−0.047 c	−0.405 c
	(0.025)	(0.222)	(0.026)	(0.223)
${CapitalFr}_{it}$	−0.039 c	−0.336 c	−0.044 c	−0.373 c
	(0.021)	(0.181)	(0.023)	(0.193)
${ForVisitFr}_{it}$	−0.016	−0.152	−0.018	−0.164
	(0.011)	(0.104)	(0.012)	(0.108)
$x{FDIOp}_{it}$	−0.038 c	−0.352 c	−0.041 c	−0.369 c
	(0.022)	(0.204)	(0.024)	(0.216)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Service} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^s$)
${ForOwnOp}_{it}$	0.529 a	3.861 a	0.540 a	4.060 a
	(0.090)	(0.728)	(0.087)	(0.722)
${CapitalFr}_{it}$	0.230 a	1.586 a	0.225 a	1.585 a
	(0.051)	(0.377)	(0.052)	(0.387)
${ForVisitFr}_{it}$	0.175 a	1.159 a	0.179 a	1.201 a
	(0.042)	(0.304)	(0.046)	(0.334)
$x{FDIOp}_{it}$	0.385 a	2.348 a	0.397 a	2.466 a
	(0.060)	(0.392)	(0.062)	(0.417)

Notes: The corresponding dependent variables are as follows: ${FEmp}_{it}^e$ (female employment-to-population ratio, %); ${ShFEmp}_{it}^a$ (share of female employment in the agricultural sector, % of female employment); ${ShFEmp}_{it}^m$ (share of female employment in the industrial sector, % of female employment); and ${ShFEmp}_{it}^s$ (share of female employment in the service sector, % of female employment). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively.

, and they conform to the benchmark empirical findings. The estimated coefficients on $x{FDIOp}_{it}$ are positive and statistically significant at the 1% level in the case of the service sector, and negative and statistically significant at the 1% level in the case of the agricultural sector. They show that a one-unit increase in $x{FDIOp}_{it}$ is associated with short-run and long-run increases in ${ShFEmp}_{it}^s$ by 0.39 pps and 2.35 pps, respectively, ceteris paribus. In contrast, a one-unit increase in this indicant would lead to short-run and long-run decreases in ${ShFEmp}_{it}^a$ by 0.31 pps and 1.84 pps, respectively. The corresponding estimates in the case of the industrial sector are found to be negative, but they are small in magnitude and exhibit statistical significance only at the 10% level. Also, consistent with the benchmark results, the effect of $x{FDIOp}_{it}$ on ${FEmp}_{it}^e$ is found to be statistically insignificant.

Next, the empirical analysis takes a closer look into the effects of FDI openness on female employment using the sub-component indicators that cover different aspects of moving international capital and foreign nationals. The results generally conform to the benchmark findings. In the case of the service sector, the estimated coefficients of the sub-component indicators are all positive and statistically significant at the 1% level, where the effect of foreign ownership/investment freedom is found to be the highest in magnitude; the short-run and the long-run effects of a one-unit increase in ${ForOwnOp}_{it}$ on ${ShFEmp}_{it}^s$ stand at increases by 0.53 pps and 3.86 pps, respectively. In the case of the agricultural sector, the estimates are all negative and statistically significant, with foreign ownership/investment freedom exercising the strongest negative effect on ${ShFEmp}_{it}^a$ among the sub-component indicators; the short-run and the long-run effects of a one-unit increase in ${ForOwnOp}_{it}$ on ${ShFEmp}_{it}^a$ stand at reductions by 0.41 pps and 2.57 pps, respectively. In the case of the industrial sector, the estimates on the sub-component indicators, ${ForOwnOp}_{it}$ and ${CapitalFr}_{it}$, are negative and statistically significant at the 10% level, but the corresponding effects on ${ShFEmp}_{it}^m$ are relatively small in magnitude. Also, the effect of ${ForOwnOp}_{it}$ on ${FEmp}_{it}^e$ is found to be positive and statistically significant at the 10% level but small in magnitude; a one-unit increase in this indicator is associated with short-run and long-run increases in ${FEmp}_{it}^e$ by 0.10 pps and 0.74 pps, respectively.

The variations in the effects of FDI openness on the share of female employment across sectors over income categories are examined when using the FDI openness indicators. The one-step and two-step GMM System estimates are presented in

Table 7. Empirical results—overall FDI openness index by income tercile.

	(i)	(ii)	(iii)	(iv)
	GMM System		GMM System [Two-Step]
	Short-Run Effect	Long-Run Effect	Short-Run Effect	Long-Run Effect
	$\mathrm{Female} \mathrm{Employment} \mathrm{Rate} ({FEmp}_{it}^e$)
$x{FDIOp}_{it}$	0.035	0.261	0.041	0.297
	(0.040)	(0.296)	(0.043)	(0.305)
$d\_x{FDIOp}_{it} \left({Trc}_1\right)$	0.085 c	0.634 c	0.080 c	0.580 c
	(0.044)	(0.325)	(0.047)	(0.333)
$d\_x{FDIOp}_{it} \left({Trc}_2\right)$	0.058	0.433	0.055	0.399
	(0.042)	(0.313)	(0.044)	(0.318)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Agricultural} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^a$)
$x{FDIOp}_{it}$	−0.193 a	−1.149 a	−0.181 a	−1.097 a
	(0.065)	(0.423)	(0.061)	(0.404)
$d\_x{FDIOp}_{it} \left({Trc}_1\right)$	−0.109 a	−0.649 a	−0.101 a	−0.612 a
	(0.033)	(0.212)	(0.032)	(0.208)
$d\_x{FDIOp}_{it} \left({Trc}_2\right)$	−0.174 a	−1.036 a	−0.166 a	−1.006 a
	(0.036)	(0.237)	(0.038)	(0.261)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Industrial} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^m$)
$x{FDIOp}_{it}$	−0.047 c	−0.376 c	−0.049 c	−0.402 c
	(0.026)	(0.206)	(0.027)	(0.218)
$d\_x{FDIOp}_{it} \left({Trc}_1\right)$	−0.005	−0.040	−0.005	−0.041
	(0.022)	(0.179)	(0.022)	(0.183)
$d\_x{FDIOp}_{it} \left({Trc}_2\right)$	−0.063 b	−0.504 b	−0.068 b	−0.557 b
	(0.030)	(0.237)	(0.032)	(0.236)
	$\mathrm{Share} \mathrm{of} \mathrm{Female} \mathrm{Employment} \mathrm{in} \mathrm{the} \mathrm{Service} \mathrm{Sec}\mathrm{tor} ({ShFEmp}_{it}^s$)
$x{FDIOp}_{it}$	0.227 a	1.455 a	0.235 a	1.546 a
	(0.049)	(0.353)	(0.052)	(0.379)
$d\_x{FDIOp}_{it} \left({Trc}_1\right)$	0.108 a	0.692 a	0.113 a	0.743 a
	(0.028)	(0.198)	(0.029)	(0.212)
$d\_x{FDIOp}_{it} \left({Trc}_2\right)$	0.190 a	1.218 a	0.196 a	1.289 a
	(0.047)	(0.329)	(0.050)	(0.356)

Notes: The corresponding dependent variables are as follows: ${FEmp}_{it}^e$ (female employment-to-population ratio, %); ${ShFEmp}_{it}^a$ (share of female employment in the agricultural sector, % of female employment); ${ShFEmp}_{it}^m$ (share of female employment in the industrial sector, % of female employment); and ${ShFEmp}_{it}^s$ (share of female employment in the service sector, % of female employment). Robust (Windmeijer-corrected) standard errors are reported in parentheses, with “a”, “b”, and “c” denoting statistical significance at the 1% level, 5% level, and 10% level, respectively. This table presents the results when specifying the effect of $x{FDIOp}_{it}$ by the RGDPC tercile categories. The third tercile (${Trc}^3$) is set as the reference, whereas the first and second terciles (${Trc}^1$ and ${Trc}^2$) are presented as deviations from the reference.

, and they are generally consistent with the benchmark empirical findings, albeit with some minor differences in the case of the industrial sector. They suggest that, at higher RGDPC levels, the impacts of FDI on the share of female employment across sectors moderately decrease (see Appendix A for the estimates of the total effects for ${Trc}^1$ and ${Trc}^2$).13

6. Conclusions

FDI is one of the main aspects of economic globalization, and it is often characterized as a promoting vector of economic growth and structural changes that impact women’s economic and social conditions. As such, it is deemed to have important effects on gender inequality and female economic participation. This paper empirically examines the short-run and the long-run effects of inward FDI on the female employment rate and the share of female employment across sectors. The corresponding hypothesis posits that inward FDI stimulates the transition of female labour from agriculture to manufacturing and, ultimately, to the service sector. Consequently, FDI openness is expected to decrease female employment in agriculture while increasing it in services. The effect on female employment in the industrial sector is less predictable, depending on economic factors and the relative flow of female labour from agriculture toward services through the industrial sector. The empirical analysis is executed through the GMM System estimator for dynamic panel models using different empirical specifications and FDI openness indicators. The main results show that the overall effects of inward FDI on the national female employment rate are not statistically significant. However, they reveal that higher levels of FDI openness lead to increases in the share of female employment in the service sector and reductions in the share of female employment in the agricultural sector. The effects of FDI openness on the share of female employment in the industrial sector are found to be statistically insignificant. These results are generally supported when running the empirical analysis using different FDI openness indicators that cover foreign ownership/investment openness, capital freedom, and ease of foreign nationals to visit. Supplementary analysis reveals some variations in the magnitude of these effects over national income categories, suggesting that the FDI effects on the share of female employment across sectors moderately decrease at higher national income levels as the transition of female labour across sectors relatively decelerates. It is worth noting that, due to data limitations, this paper does not account for the heterogeneous nature of FDI (e.g., greenfield investments versus mergers and acquisitions) and female labour in the informal sector. Also, external shocks, societal dynamics, and policy changes tend to alter FDI’s effects over time, rendering long-term patterns difficult to capture.

These findings are basically consistent with the prevalence of macroeconomic channels through which inward FDI impacts female employment across sectors. As such, they signify the promoting effects of inward FDI on economic development and restructuring and on the transition of women from the agricultural sector—through low-skill industrial employment—toward the service sector, which is often characterized by more cognitive, administrative, and managerial occupations. The estimates also embody the implications of various counteracting microeconomic channels with varying significance across sectors. For instance, foreign affiliates of MNEs tend to benefit from the prevalence of the gender wage gap in host countries, leading to higher female employment rates, particularly in the industrial sector in low-income countries. In contrast, they could bring in capital-intensive production processes and advanced technologies that disproportionally replace unskilled female workers. Also, foreign affiliates of MNEs normally feature cultural and business norms that encourage gender equality in the workplace. As such, they stimulate increases in female employment, particularly in the service sector in emerging economies and catalyze spillovers of these norms into domestic firms.

FDI openness policies are significant in raising female employment in the service sector, which often comprises positions and occupations that align with women’s comparative advantage. These policies, which typically aim at reducing restrictions on foreign ownership of capital, operation of foreign affiliates of MNEs, and cross-border movements of foreign nationals, would foster women’s economic transition across sectors—from agriculture toward the service sector through the industrial sector—over the process of economic development. These policies could be combined with economic and social development schemes that raise women’s well-being, social status, and economic opportunities in rural areas to sustain their economic and social roles and mitigate their migration out of the agricultural sector. The favourable effects of inward FDI on female employment could be generally enhanced through female educational and training programs that confer employment competitiveness to women, public and judicial reforms that support women’s social status, gender-inclusive business strategies that uphold corporate social responsibility, and economic policies that facilitate women’s accessibility to employment and resources.