## wp1825

## Source details

**Canonical URL:** [wp1825](https://www.imf.org/-/media/files/publications/wp/2018/wp1825.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1825.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1825.pdf.json)

---

### Introduction and motivation
- Girls and women in developing countries face limitations: lower educational enrolment and completion rates than boys; lower political participation; higher exposure to violence; greater share of time allocated to household chores; lower labor market participation rates (United Nations, 2015).
- Christine Lagarde (IMF Managing Director) quoted: “Globally, only 55 percent of women have the opportunity to participate in the labor force, compared with 80 percent for men. Women still earn about 50 percent less than men for the same type of work, and they represent only 20 percent of parliamentarians across the world”.
- Improving women’s welfare can lead to a more efficient allocation of household resources and increase expenditure shares on health, education, and nutrition (Chen, 2006; Björkman Nyqvist and Jayachandran, 2017; Castilla and Walker, 2013).

### Transmission channels: how FDI may affect gender outcomes
- Positive channels:
  - FDI expands firms and government revenue, raising labor demand and enabling investment in public facilities (e.g., public schools, medical centers, water and electricity distribution) (Braunstein, 2006).
  - Sectoral composition: FDI into female-intensive sectors increases female labor demand (Aguayo-Tellez, et al., 2010).
  - Technological spillovers from MNEs can raise wages via skill upgrading; if technology favors female workers, female labor demand may rise (Glass and Saggi, 2002; World Bank, 2012).
  - Corporate Social Responsibility (CSR) and practices of majority-foreign owned firms can promote gender-equal norms (Abe, Javorcik, and Kodama, 2016; Olcott and Oliver, 2014; Kucera, 2001).
- Negative channels:
  - Cultural norms may block beneficial spillovers (Kwok and Tadesse, 2006).
  - FDI concentrated in technologically advanced sectors may disadvantage women (Tejani and Milberg, 2010).

### Novel contributions of this study
- Uses two composite UNDP gender indices: the Gender Inequality Index (GII) and the Gender Development Index (GDI).
- Examines both gender development and gender inequality dimensions and decomposes indices into subcomponents.
- First assessment (to authors’ knowledge) of whether specific gender policies alter the impact of FDI on gender development and inequality:
  - equal wage law,
  - non-discrimination in hiring law,
  - gender budgeting programs,
  - female access to resources (interacted with FDI inflows).

### Data, sample, and empirical approach
- Sample: 94 developing countries over 1990–2015.
  - Regional counts: Sub-Saharan Africa 32; Latin America and Caribbean 19; Europe and Central Asia 15; Middle-East and North Africa 11; South East Asia 6; East Asia and Pacific 11.
  - Income classification: 69 countries MIC; 25 countries LIC (IMF WEO classification).
- Dependent variables:
  - GDI: index 0 to 1 (higher = women better off).
  - GII: index 0 to 1 (higher = increased disparities).
  - Both indices available for 188 countries, 1990–2015 (UNDP).
- Main explanatory variable: FDI inflows per capita (IMF Balance of Payments database). Robustness also uses FDI % of GDP and alternative sources (IMF WEO, World Bank, UNCTAD).
- Estimation:
  - System of simultaneous linear equations for GDI and GII estimated by Seemingly Unrelated Regression Equations (FGLS) with country and year fixed effects.
  - Controls X′ include: trade openness, GDP growth, Polity2, government expenditures (% of GDP), natural resources rents (% of GDP), government gross debt (% of GDP), rural population (%), female population (%).
  - Interaction analysis with gender policy variables and access-to-resources index.
  - Robustness: alternative FDI sources, IV approach, additional cultural controls, alternative econometric methods.

### Main empirical findings — baseline associations and magnitudes
- Core associations:
  - FDI inflows per capita positively associated with GDI and negatively associated with GII.
- Baseline coefficients (examples from Table 1):
  - Log(FDI inflows): 0.00372*** (GDI, std. err. 0.0005).
  - Log(FDI inflows): -0.00243** (GII, std. err. 0.0009).
  - Lagged Log(FDI inflows), t-1: 0.00348*** (GDI) and -0.00210** (GII) with std. errs. (0.000538) and (0.000954).
  - Statistical significance: *** p<0.01; ** p<0.05; * p<0.1.
- Other significant baseline control relationships (selected):
  - Log(trade openness): 0.00879*** (GDI, std. err. 0.0029).
  - Polity2: 0.00101*** (GDI); -0.000812* (GII).
  - Log(public debt): -0.00532*** (GDI); 0.00491** (GII).
  - Female population (%): 0.00388** (GDI); -0.0110*** (GII).
  - Government expenditure: 0.000658*** (GDI).
  - Natural resources rents: 0.000875*** (GII).
- Goodness of fit: R-squared values around 0.96–0.97 in many specifications.

### Heterogeneity — income and regional differences
- Income groups:
  - FDI effects on gender development stronger in middle-income countries than in low-income countries.
  - Effects on gender inequality similar across low- and middle-income subgroups.
- Selected regional Log(FDI inflows) coefficients (Table 3):
  - SSA: 0.00329*** (GDI) and 0.000935 (GII).
  - Latin America & Caribbean: 0.00230** (GDI) and -0.000588 (GII).
  - Europe & Central Asia: 0.00253* (GDI) and -0.0116** (GII).
  - Middle East & North Africa: 0.00166** (GDI) and -0.00677* (GII).
  - South Asia: -0.00744* (GDI) and 0.0060 (GII).
  - East Asia & Pacific: 0.00212*** (GDI) and -0.00388* (GII).
- Interpretation:
  - FDI positive and significant for GDI in most regions except South East Asia (negative in some specifications).
  - FDI negative and significant for GII (reducing inequality) mainly in Europe & Central Asia, Middle-East & North Africa, and East Asia & Pacific.

### GDI and GII subcomponents — channels and magnitudes
- GDI subcomponents (selected coefficients, Table 4):
  - Female life expectancy: Log(FDI inflows) 0.299*** (std. err. 0.0563).
  - Years of schooling, female: Log(FDI inflows) 0.779*** (0.0374).
  - GNI per capita (female): Log(FDI inflows) 0.341*** (0.0108).
  - Female-to-male years of schooling ratio: Log(FDI inflows) 0.0515*** (0.0032) to 0.0331*** (0.0037).
- GII subcomponents (selected coefficients, Table 5):
  - Maternity mortality: Log(FDI inflows) -7.063*** (1.7080).
  - Gross enrolment secondary level (female): Log(FDI inflows) 0.00874*** (0.00157).
  - Female parliamentary seats: Log(FDI inflows) -0.271* (0.1430); -0.307** (0.1420) in alternative specification.
  - Female-to-male labor participation ratio: not statistically significant in main specifications (e.g., -0.0608 (0.0903); -0.00385 (0.0873)).
- Channel interpretation:
  - Health: reductions in maternity mortality and increases in female life expectancy.
  - Education: increases in female schooling and secondary enrolment parity.
  - Income: increases in female GNI per capita.
  - Political empowerment and labor participation show mixed or limited effects.

### Employment and labor-market dimensions (Table 6)
- Selected associations of Log(FDI inflows) with employment variables:
  - Female informal employment (% of total employment): -5.137*** (std. err. 1.0850).
  - Female-to-male part-time employment ratio (% of total employment): -0.0974** (0.0485).
  - Gender wage gap (25 years avg): -2.974*** (0.4560).
  - Female-to-male employers ratio: 0.0103 (not significant).
- Interpretation:
  - FDI associated with substantial decreases in female informal employment, reductions in female part-time share relative to men, and reductions in the gender wage gap.

### Interaction with gender policies and access to resources (Table 7)
- Interacted policy variables mostly not statistically significant for improving FDI effects (selected interactions):
  - Log(FDI inflows)*Gender budgeting: -0.00040 (GDI); 0.000517 (GII) — not significant.
  - Log(FDI inflows)*Equal wage law: -0.001580 (GDI); 0.00429 (GII) — not significant.
  - Log(FDI inflows)*Non-discrimination law: -0.000729 (GDI); 0.00113 (GII) — not significant.
- Access to resources and procedural barriers:
  - Log(FDI inflows)*Low access to resources: -0.00377 (GDI); 0.0104** (GII, std. err. 0.0051) — interaction with GII positive and significant, indicating FDI increases gender inequality where female access to resources is low.
  - Log(FDI inflows)*number of procedures to open a business (women face higher number): -0.0019** (GDI, std. err. 0.0009) and 0.0041** (GII, std. err. 0.0018) — significant interactions indicate:
    - The positive impact of FDI on gender development is dampened where women face higher procedural burdens.
    - FDI may increase gender inequality where women face more procedures to open businesses.
- Net marginal effect: poor female access to resources and higher procedural burdens for women reduce or reverse positive gender outcomes from FDI.

### Robustness tests (selected)
- Alternative FDI sources (IMF WEO, UN): coefficients remain nearly identical to baseline (Appendix IV). Example ranges:
  - Log(FDI inflows): 0.00314***; -0.00372***; 0.00275***; -0.00327*** (GDI / GII columns).
- Instrumental variables (U.S. equity-market uncertainty index instrument): yields similar positive association with GDI and negative with GII (Appendix IV, Table 9).
- Cultural and norms controls (ethnic fractionalization; religion shares): main findings remain intact (Appendix VI). Religion associations noted:
  - Higher share of Muslims and Hinduists associated with lower GDI and higher GII; Catholics and Protestants positively associated with GDI (coefficients and significances in Appendix VI).
- Alternative FDI measure: FDI inflows as % of GDP — results consistent (Appendix VII).
- Alternative dependent indices (Stotsky et al. 2016): results in line with baseline (Appendix VIII).

### Summary statistics (selected exact values)
- Log(FDI inflows) IMF STA database: Obs 1663; Mean 3.6; Std. Dev. 1.9; Min -7.3; Max 10.8.
- GDI UNDP: Obs 2031; Mean 0.9; Std. Dev. 0.1; Min 0.5; Max 1.1.
- GII UNDP: Obs 1844; Mean 0.5; Std. Dev. 0.1; Min 0.1; Max 0.8.
- Female life expectancy: Obs 1985; Mean 66.8; Std. Dev. 10.2; Min 30.0; Max 82.1.
- Maternity mortality ratio (per 100,000 live births): Obs 2031; Mean 322.4; Std. Dev. 363.0; Min 7.0; Max 2900.0.
- Female informal employment (in percentage of total employment): Obs 766; Mean 55.9; Std. Dev. 21.4; Min 4.2; Max 89.2.
- Gender wage gap International Labour Organization: Obs 962; Mean 16.6; Std. Dev. 14.3; Min -27.7; Max 47.2.
- Gender budgeting dummy: Obs 2021; Mean 0.3; Std. Dev. 0.4; Min 0.0; Max 1.0.
- Female access to resources (OECD): Obs 1945; Mean 0.4; Std. Dev. 0.2; Min 0.0; Max 1.0.

### Conclusions — core conclusions and mechanisms
- Core empirical conclusion:
  - FDI inflows per capita are positively associated with gender development (GDI) and negatively associated with gender inequality (GII).
- Main channels identified:
  - Health: increases in female life expectancy; decreases in maternal mortality.
  - Education: increases in female years of schooling and female-to-male gross enrolment in secondary schools.
  - Income & labor: increases in female GNI per capita; reductions in female informal employment; reductions in gender wage gap; decreases in female part-time employment share relative to men.
- Heterogeneity:
  - Stronger positive impact on gender development in Sub-Saharan Africa, Latin America & Caribbean, Middle East & North Africa, and East Asia & Pacific.
  - Negative association with gender development in South East Asia (attributed to FDI into technological, male-skewed sectors).
  - Stronger impact on gender development in middle-income than low-income countries.
- Policy-modulating factors:
  - Low female access to resources (land, non-land assets, financial services) can lead to FDI increasing gender inequality.
  - Procedural burdens to open businesses for women weaken or reverse FDI’s positive effects on gender development.

### Policy recommendations (as drawn from the paper)
- Promote foreign investments with attention to sectoral composition and labor intensity to maximize female labor demand.
- Encourage corporate social responsibility (CSR) initiatives and workplace practices promoting gender equality (e.g., non-discrimination, family-friendly arrangements).
- Improve women’s access to resources (land, assets, financial services) so women can capture and control income gains from FDI.
- Simplify and shorten procedures and paperwork for women to open businesses to enhance women’s entrepreneurial opportunities from FDI.
- Pursue sectoral and firm-level analyses in future research to identify where FDI produces the largest gender gains.

*Source: wp1825 (IMF Working Paper; Sections II–III and supporting tables and appendixes as provided).*

### References .............................................................................................................

### wp1825 - References

### Introduction and motivation
- Girls and women in developing countries face limitations: lower educational enrolment and completion rates than boys; lower political participation; higher exposure to violence; greater share of time allocated to household chores; lower labor market participation rates (United Nations, 2015).
- Christine Lagarde (IMF Managing Director) quoted: “Globally, only 55 percent of women have the opportunity to participate in the labor force, compared with 80 percent for men. Women still earn about 50 percent less than men for the same type of work, and they represent only 20 percent of parliamentarians across the world”.
- Improving women’s welfare can lead to a more efficient allocation of household resources and increase expenditure shares on health, education, and nutrition (Chen, 2006; Björkman Nyqvist and Jayachandran, 2017; Castilla and Walker, 2013).

### Transmission channels: how FDI may affect gender outcomes
- FDI can expand firms and increase government revenue, raising labor demand and enabling investment in public facilities and infrastructure that empower women (e.g., public schools, medical centers, water and electricity distribution) (Braunstein, 2006).
- Sectoral composition: if FDI flows into sectors that rely proportionally more on a female workforce, female labor demand increases relative to male demand (Aguayo-Tellez, et al., 2010).
- Technological spillovers from MNEs and majority-foreign owned firms can raise wages via skill upgrading in local firms; if technology favors female workers, female labor demand may rise (Glass and Saggi, 2002; World Bank’s World Development Report, 2012).
- Corporate Social Responsibility (CSR) initiatives and foreign investors can promote gender-equal norms (Abe, Javorcik, and Kodama, 2016; Olcott and Oliver, 2014; Kucera, 2001).
- Potential negative effects: FDI could worsen gender outcomes if national cultural norms oppose female development (Kwok and Tadesse, 2006) or if investments are in technologically advanced sectors that disadvantage women (Tejani and Milberg, 2010).

### Novel contributions of this study
- Uses two composite UNDP gender indices rather than single indicators: the Gender Inequality Index (GII) and the Gender Development Index (GDI).
- Examines both gender development and gender inequality dimensions; previous literature often focused on a single outcome (e.g., labor force participation, wage gap, or years of schooling).
- First assessment (to the extent of authors’ knowledge) of whether specific gender policies alter the impact of FDI on gender development and inequality, including:
  - equal wage law,
  - non-discrimination in hiring law,
  - gender budgeting programs,
  - female access to resources (interacted with FDI inflows).

### Data, sample, and empirical approach
- Panel covers 94 developing countries over the period 1990–2015.
- Estimation method: Seemingly Unrelated Regressions Equations (to control for potential correlation between GDI and GII).
- Decomposes GDI and GII and estimates equations for each subcomponent; examines employment-related gender variables and policy interactions.

### Main empirical findings
- FDI inflows are positively associated with gender development (women are better off) and negatively associated with gender inequality (decreasing gender disparities).
- Effects are stronger and more significant for middle income countries and specific geographic regions, including Sub-Saharan Africa, Latin America, and Middle East and North Africa.
- Subcomponent effects:
  - FDI inflows particularly affect women’s life expectancy and maternity mortality ratio, and the female-to-male ratio of gross enrolment rate to the secondary level.
  - Regarding employment, FDI inflows are negatively associated with female informal employment; negatively associated with female-to-male ratio of part-time employment in percentage of total employment; and negatively associated with the gender wage gap.
- Interaction and conditional effects:
  - FDI inflows increase gender disparities if women have restricted access to land and non-land assets, and to financial resources.
  - The impact of FDI on gender issues is reduced in countries where the number of procedures to open a business is higher for women than for men.
- Policy implication drawn: to fully benefit from FDI inflows, countries need to improve the business environment for women and lift barriers to access resources so women can enjoy free access to the labor market and new income opportunities.

### Organization of the paper (as presented)
- Section II provides a brief overview of the literature.
- Additional materials included in the source: Tables 1–12, Figures 1–3, Appendixes I–VII, and a list of countries and summary statistics.

*Source: wp1825 - References (IMF Working Paper; content covers Introduction and overview of findings and methodology).*

### Section III presents our data and exposes our empirical methodology. Section IV highlights

### wp1825 - Section III presents our data and exposes our empirical methodology. Section IV highlights

### III. Data and Empirical Methodology — Data Sources
- Sample:
  - 94 developing countries from 1990–2015.
  - Regional coverage: Sub-Saharan Africa (32 countries), Latin America and Caribbean (19), Europe and Central Asia (15), Middle-East and North Africa (11), South East Asia (6), East Asia and Pacific (11).
  - Income classification: 69 countries are Lower or Upper Middle Income (MIC) and 25 are Low Income countries (LIC) according to IMF’s World Economic Outlook classification.
- Dependent variables:
  - Gender Development Index (GDI): index ranging from 0 to 1, higher values indicate women are better off; measures disparities between women and men in health (life expectancy), knowledge (expected and mean years of schooling), and living standards (GNI per capita). Computation explained in appendix.
  - Gender Inequality Index (GII): scale from 0 to 1, higher values indicate increased disparities between women and men; measures reproductive health (maternal mortality ratio and adolescent fertility rate), empowerment (female share of parliamentary seats and educational attainment at secondary level), and labor participation rate (women and men aged 15 or older).
  - Both indices available for 188 countries from 1990–2015 (data extracted from the United Nations Development Program).
- Main explanatory variable:
  - FDI inflows per capita extracted from IMF’s Balance of Payments database.
  - Rationale for per capita measure: avoids potential endogeneity between FDI and GDP (FDI could increase GDP, and vice-versa); population is more stable so per capita captures real dynamics rather than GDP fluctuations.
  - Robustness: will also use FDI in percentage of GDP and alternative data sources (IMF’s World Economic Outlook, World Bank’s World Development Indicators, UNCTAD).
- Control variables:
  - Data sources include IMF’s WEO database; the United Nations; UNCTAD; the World Bank’s database; the World Bank’s Gender statistics database; ILO; UNESCO; Polity IV data series; and the OECD.
- Additional methodological notes:
  - Use of two UNDP indices (GDI and GII) and their subcomponents to assess both gender development and gender inequality rather than single indicators used in prior literature.
  - Interaction analysis with gender policies: equal wage law, non-discrimination in hiring law, gender budgeting programs, and female access to resources to assess whether these policies alter FDI’s impact on gender outcomes.
  - Robustness checks include multiple FDI data sources and an alternative econometric method to address endogeneity.

### II. Literature Review — Theoretical Insights
- Ambiguity of FDI effects on gender:
  - Positive channels:
    - General equilibrium effect: FDI expands financial capital, brings new technology, improves export potential, increases government revenue; reinvestment can improve public facilities (e.g., school construction) and reduce gender educational gaps (Braunstein, 2006).
    - Sectoral composition: If FDI is concentrated in female-intensive sectors, female labor demand increases in absolute and relative terms; explained via Hecksher-Ohlin specialization of developing countries in low-skilled, labor-intensive sectors.
    - Becker (1971) argument: increased competition makes discrimination more costly and may reduce the gender wage gap.
    - Technological spillovers: MNEs’ vertical relationships with local firms may induce technology adoption and worker training, potentially increasing female labor demand if technologies are complementary to female workforce (Acemoglu, 1998) and offering wage premiums/job security (Glass and Saggi, 2002).
    - Corporate Social Responsibility (CSR): MNEs may implement CSR initiatives improving working conditions, health and safety (OECD, 2001); majority-owned foreign firms more likely to have higher share of female directors, family-friendly arrangements, and lower gender wage gaps (Kucera, 2001; Olcott and Oliver, 2014; Abe, Javorcik and Kodama, 2016); spillovers to local firms (UNCTAD, 2014; Carr, 2016).
  - Negative channels and caveats:
    - FDI can increase labor demand elasticity and decrease job security for women.
    - Cultural and social norms, including religion, may block spillovers of firm practices into broader society (Kwok and Tadesse, 2006; Carr, 2016; Cooray and Potrafke, 2011).
    - Heterogeneity by stage of industrialization: early industrialization attracts low-skill, female-intensive manufacturing; later stages attract skilled, often male-dominated labor, creating potential anti-female bias (Tejani and Milberg, 2010; Vijaya and Kaltani, 2007).
- Three main transmission channels identified in literature:
  - General Equilibrium Effect
  - Technological Spillovers
  - Corporate Social Responsibility (CSR)

### II. Literature Review — Empirical Findings
- Outcome measurement heterogeneity:
  - Most empirical studies use singular indicators (gender wage gap or labor force participation rate gap), producing mixed conclusions.
- Evidence on gender wage gap:
  - Rasekhi and Hosseinmardi (2012): panel regression, sample of 21 developing countries, 2000–2007, found FDI inflows reduced the gender wage gap but warned of potential widening if firms exploit lower female wages.
  - Vijaya and Kaltani (2007): fixed-effects panel of 19 countries, 1987–2001, found FDI inflows and FDI stocks in manufacturing had a negative impact on female workers’ wages.
  - Davin (2004): descriptive evidence from Chinese EPZs showing women in export factories earned a month’s pay exceeding a year’s pay for male village workers.
- Evidence on labor force participation:
  - Argument that FDI generates employment opportunities for women because labor-intensive industries absorb female workers; foreign firms pursuing cost advantages create factory jobs for women (Seguino and Grown, 2006).
  - Countervailing forces: increased competition and capital mobility could make domestic labor demand more elastic and drive cost-cutting, harming female employment prospects (Rodrick, 1997; Seguino and Grown, 2006).
  - Jaffri and others (2015): FDI in technological and service sectors ambiguous; FDI in non-services sectors positively affects female labor force participation.
- Evidence on CSR / foreign ownership effects:
  - Chen and others (2013): Chinese firm census (2004) found foreign participation and export orientation increase female workforce by 13 percent compared to non-export sectors, reduce the gender wage gap, and show no significant gender wage discrimination.
  - Studies in Japan (Abe, Javorcik, and Kodama, 2016; Carr, 2016) find majority-owned foreign firms more likely to have a higher share of female directors, lower gender wage gap, and family-friendly arrangements.
- Macro evidence on welfare:
  - Anyanwu (2016): cross-sectional time series OLS, 1991–2011, found a 1 percent growth in FDI in percentage of GDP increases gender equality in youth employment (ratio of female to male employment for age 15–24) by 0.55 percentage point in all of Africa and by 0.44 percentage point in sub-Saharan Africa (SSA).
- Contributions of the present study relative to prior work:
  - Use of two UNDP indices (GDI and GII) and their subcomponents rather than single indicators.
  - Simultaneous analysis of gender development and gender inequality.
  - First (to authors’ knowledge) to assess whether equal wage law, non-discrimination in hiring law, gender budgeting programs, and female access to resources alter the impact of FDI on gender outcomes via interaction analysis.
  - Extensive robustness tests with multiple FDI sources and an alternative econometric method to address endogeneity.

### III. Stylized Facts (selected empirical observations)
- Visual relationships (Figures referenced):
  - Figure 1: FDI inflows per capita co-move with GDI and GII in developing countries over 1990–2015.
    - High levels of FDI inflows per capita appear associated with higher GDI values and lower GII values.
    - Reported statistics for Figure 1:
      - GDI panel: Correlation coefficient: 0.5658        R-squared: 0.3201
      - GII panel: Correlation coefficient: -0.6964          R-squared: 0.4850
  - Figure 2 (decomposed by income):
    - FDI inflows per capita in middle-income countries seem more strongly positively related to GDI and more strongly negatively related to GII than in low-income countries.
    - Interpretation: impact of FDI on women’s welfare and gender disparities appears larger in middle-income countries; possible explanations include differences in size of FDI inflows, technology, and absorption capacity, and sectoral composition (services and manufacturing vs. primary sector).
  - Figure 3 (geographic changes 1990–2015):
    - Biggest increase in FDI inflows occurred in Europe and Central Asia, Sub-Saharan Africa, and South East Asia.
    - South East Asia and Middle-East and North Africa experienced the biggest improvement in both women’s welfare (GDI) and the biggest decrease in gender disparities (GII):
      - South East Asia’s GDI increased by 0.15 points.
      - Middle-East and North Africa’s GDI increased by 0.13 points.
      - South East Asia’s GII decreased by [value not provided in supplied excerpt].
- Note: Figures and some numeric details are presented in the original figures referenced in the text.

_Italic: Source — wp1825, Sections II–III as provided in the supplied PDF content._

### 0.20 points and Middle-East and North Africa’s by-0.23).

### wp1825 - 0.20 points and Middle-East and North Africa’s by-0.23).

### Empirical Methodology
- System of simultaneous linear equations estimated for GDI and GII:
  - Equation (1): GDI_i,t = ∝1 + β1 FDI_i,t + θ1 X′_i,t + v1_i + φ1_t + ε_i,t
  - Equation (2): GII_i,t = ∝2 + β2 FDI_i,t + θ2 X′_i,t + v2_i + φ2_t + u_i,t
- FDI: foreign direct investment inflows per capita. GDI: Gender Development Index. GII: Gender Inequality Index.
- Fixed effects: country (v1,2_i) and year (φ1,2_t).
- Estimation method: Seemingly Unrelated Regression Equations (Feasible Generalized Least Squares - FGLS) — OLS residuals used to estimate full variance-covariance matrix; weights applied by covariance of residuals for efficiency and joint-equation testing.
- Controls (X′_i,t) included:
  - Trade openness = trade (exports + imports) to GDP ratio.
  - GDP growth.
  - Polity 2 (Polity IV).
  - General government total expenditures (% of GDP).
  - Total natural resources rents (% of GDP).
  - General government gross debt (% of GDP).
  - Rural population (% of total population).
  - Female population (% of total population).

### Baseline Results — Main Findings
- Overall associations:
  - FDI inflows per capita are positively associated with GDI and negatively associated with GII.
  - Baseline coefficient examples (Table 1):
    - Log(FDI inflows): 0.00372*** (GDI, column 1) with standard error (0.0005).
    - Log(FDI inflows): -0.00243** (GII, column 2) with standard error (0.0009).
    - Log(FDI inflows), t-1: 0.00348*** (GDI) and -0.00210** (GII) with standard errors (0.000538) and (0.000954) respectively.
  - Statistical significance legend used: *** p<0.01; ** p<0.05; * p<0.1.
- Other significant baseline relationships (Table 1):
  - Log(trade openness) positively associated with GDI (e.g., 0.00879***, (0.0029)); not significant for GII in many specifications.
  - Polity2 positively associated with GDI and negatively with GII (e.g., 0.00101*** for GDI; -0.000812* for GII).
  - Log(public debt) negatively associated with GDI and positively with GII (e.g., -0.00532*** GDI; 0.00491** GII).
  - Female population (%): positive with GDI and negative with GII (e.g., 0.00388** GDI; -0.0110*** GII).
  - Government expenditure: positive with GDI (0.000658***).
  - Natural resources rents: positive with GII (0.000875***).
  - GDP growth and rural population coefficients: not statistically significant in baseline.
- Goodness of fit examples: R-squared values around 0.96–0.97 in many specifications (Table 1).

### Heterogeneity — Income and Geographic Subgroups
- Income subgroups (Table 2):
  - FDI effects on gender development stronger in middle-income countries than in low-income countries.
  - Effects on gender inequality similar across low- and middle-income subgroups.
- Geographic subgroups (Table 3) — selected coefficients for Log(FDI inflows) by region:
  - SSA (Sub-Saharan Africa): 0.00329*** (GDI) and 0.000935 (GII).
  - Latin America & Caribbean: 0.00230** (GDI) and -0.000588 (GII).
  - Europe & Central Asia: 0.00253* (GDI) and -0.0116** (GII).
  - Middle East & North Africa: 0.00166** (GDI) and -0.00677* (GII).
  - South Asia: -0.00744* (GDI) and 0.0060 (GII)  [note: South-East Asia described as negative elsewhere].
  - East Asia & Pacific: 0.00212*** (GDI) and -0.00388* (GII).
- Interpretation: FDI positive and significant for GDI in most regions except South-East Asia (negative). FDI negative and significant for GII (reducing inequality) mainly in Europe & Central Asia, Middle-East & North Africa, and East Asia & Pacific.

### GDI and GII Subcomponents — Channels
- GDI subcomponents (Table 4) — FDI associations (selected):
  - Female life expectancy: Log(FDI inflows) 0.299*** (std. err. 0.0563).
  - Years of schooling, female: Log(FDI inflows) 0.779*** (0.0374).
  - GNI per capita (female): Log(FDI inflows) 0.341*** (0.0108).
  - Female-to-male life expectancy ratio: Log(FDI inflows) 0.00040 (0.0003) to 0.0005* (0.0003) depending on specification.
  - Female-to-male years of schooling: Log(FDI inflows) 0.0515*** (0.0032) to 0.0331*** (0.0037).
- GII subcomponents (Table 5) — FDI associations (selected):
  - Maternity mortality: Log(FDI inflows) -7.063*** (1.7080).
  - Gross enrolment secondary level (female): Log(FDI inflows) 0.00874*** (0.00157).
  - Female parliamentary seats: Log(FDI inflows) -0.271* (0.1430) in one specification; -0.307** (0.1420) in another.
  - Female-to-male labor participation ratio: not statistically significant in main specifications (e.g., -0.0608 (0.0903); -0.00385 (0.0873)).
- Summary channel interpretation:
  - FDI improves women’s health (lower maternal mortality, higher life expectancy) and education (higher female schooling and enrolment).
  - Effects on political empowerment (female parliamentary seats) and labor participation ratios are mixed or not significant.
  - FDI associated with increased female GNI per capita.

### Other Employment Variables (Table 6) — FDI and Labor Market Dimensions
- Employment indicators analyzed:
  - Female-to-male employers ratio (percentage of employment).
  - Female informal employment (% of total employment, 25 years avg).
  - Female-to-male part-time employment ratio (% of total employment, 25 years avg).
  - Gender wage gap (25 years avg).
- Selected FDI associations (Table 6):
  - Log(FDI inflows): 0.0103 (Female-to-male employers ratio) — not significant.
  - Log(FDI inflows): -5.137*** (Female informal employment) with std. err. (1.0850).
  - Log(FDI inflows): -0.0974** (Female-to-male part-time employment ratio) (0.0485).
  - Log(FDI inflows): -2.974*** (Gender wage gap) (0.4560).
- Interpretation:
  - FDI correlated with decreases in female informal employment, decreases in female part-time share relative to men (suggesting more full-time positions for women), and reductions in the gender wage gap.

### FDI Inflows and Gender Policies (Table 7) — Interaction Results
- Policy variables tested via interaction with Log(FDI inflows):
  - Gender budgeting (dummy).
  - Law mandating equal remuneration (dummy).
  - Law mandating non-discrimination in hiring (dummy).
  - Female access to resources index (OECD; 0 to 1 with 0 = no discrimination).
  - Number of procedures to register a business for women relative to men (dummy if women face higher number).
- Interaction findings (selected):
  - Log(FDI inflows)*Gender budgeting: coefficient -0.00040 (GDI) and 0.000517 (GII) — not statistically significant.
  - Log(FDI inflows)*Equal wage: -0.001580 (GDI) and 0.00429 (GII) — not statistically significant.
  - Log(FDI inflows)*Non-discrimination: -0.000729 (GDI) and 0.00113 (GII) — not statistically significant.
  - Log(FDI inflows)*Low access to resources: -0.00377 (GDI) and 0.0104** (GII, std. err. 0.0051) — interaction with GII positive and significant (2nd-order effect increases gender inequality where female access to resources is low).
  - Log(FDI inflows)*number of procedures to open a business: -0.0019** (GDI, std. err. 0.0009) and 0.0041** (GII, std. err. 0.0018) — significant interactions indicate:
    - Negative interaction with GDI: the impact of FDI on gender development is dampened where women face higher procedural burdens.
    - Positive interaction with GII: FDI may increase gender inequality where women face more procedures to open businesses.
- Net marginal effects noted: poor access to resources and higher procedural burdens for women reduce or reverse positive gender outcomes from FDI.

### Robustness Tests
- Alternate FDI data sources: United Nations datasets and IMF’s World Economic Outlook — results nearly identical to baseline (Table 8, Appendix III).
- Instrumental variables: FDI instrumented with U.S. equity-market uncertainty index (Ouedraogo, 2017 approach) — Table 9 (Appendix IV) yields similar positive association with GDI and negative with GII.
- Controls for norms and social culture: ethnic fractionalization and religion shares (Islam, Catholicism, Protestantism, Hinduism, Buddhism) included (Table 10) — main findings remain intact. Additional findings: higher share of Muslims and Hinduists associated with lower GDI and higher GII; shares of Catholics and Protestants positively associated with GDI.
- Alternative FDI measure: FDI inflows as % of GDP (Table 11, Appendix VI) — results consistent with baseline.
- Alternative dependent variables: indices from Stotsky and others (2016) (Table 12, Appendix VII) — results in line with baseline.

### Conclusions — Core Conclusions and Mechanisms
- Core empirical conclusion:
  - FDI inflows per capita are positively associated with gender development (GDI) and negatively associated with gender inequality (GII).
- Main channels:
  - Health: increases in female life expectancy; decreases in maternal mortality.
  - Education: increases in female years of schooling and female-to-male gross enrolment in secondary schools.
  - Income & labor: increases in female GNI per capita; reductions in female informal employment; reductions in gender wage gap; decreases in female part-time employment share relative to men (suggesting more full-time female employment).
- Heterogeneity:
  - Stronger positive impact on gender development in Sub-Saharan Africa, Latin America & Caribbean, Middle East & North Africa, and East Asia & Pacific.
  - Negative association with gender development in South East Asia (attributed to FDI into technological, male-skewed sectors).
  - Stronger impact on gender development in middle-income than low-income countries.
- Policy-modulating factors:
  - Low female access to resources (land, non-land assets, financial services) can lead to FDI increasing gender inequality.
  - Burdensome procedures to open businesses for women weaken or reverse FDI’s positive effects on gender development.

### Policy Recommendations (as drawn from the paper)
- Promote foreign investments in developing countries, with attention to sectoral composition and labor intensity to maximize female labor demand.
- Encourage corporate social responsibility (CSR) initiatives and workplace practices that promote gender equality (e.g., non-discrimination, family-friendly arrangements).
- Improve women’s access to resources (land, assets, financial services) so that women can capture and control income gains from FDI.
- Simplify and shorten procedures and paperwork for women to open businesses to enhance women’s ability to seize entrepreneurial opportunities stemming from FDI.
- Recognize need for sectoral/firm-level analyses in future research to identify where FDI produces the largest gender gains.

*Source: IMF working paper wp1825 (text and tables as provided).*

### REFERENCES

### REFERENCES

### Key References (selected from list)
- Abe, Y., B. Javorcik, and N. Kodama, 2016, Transplanting Corporate Culture across International Borders: FDI and female employment in Japan, RIETI Discussion Paper Series, 16-E-015.
- Acemoglu, D., 1998, Why Do New Technologies Complement Skills? Directed Technical Change and Wage Inequality, The Quarterly Journal of Economics, Vol. 113, No. 4, pp. 1055–1089.
- Aguayo-Tellez, E., and Airola, J. and Juhn, C., 2010, Did Trade Liberalization Help Women? The Case of Mexico in the 1990s, National Bureau of Economic Research, Inc, NBER Working Papers: 16195.
- Alfaro.L, Kalemli-Ozcan.S, and V. Volosovych, 2008, Why doesn't Capital Flow from Rich to Poor Countries? An Empirical Investigation, Review of Economics and Statistics, May, 90(2): pp. 347–368.
- Becker, G.S., 1971, “The economics of discrimination,” University of Chicago Press, second edition.
- Braunstein, E., 2006, “Foreign Direct Investment, Development and Gender Equity: A Review of Research and Policy,” United Nations Research Institute for Social Development, occasional paper.
- Javorcik, B. S., 2004, “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers Through Backward Linkages,” American Economic Review, Vol. 94, No. 3, pp. 605–627.
- Kazandjian, R., L. Kolovich, K. Kochhar, and M. Newiak, 2016, “Gender Equality and Economic Diversification,” IMF Working Paper 16/140, Washington: International Monetary Fund
- Kochhar, K., S. Jain-Chandra, and M. Newiak, 2016, Women, Work, and Economic Growth: Leveling the Playing Field, eds 2016. International Monetary Fund, Washington, DC.
- Neumayer, E., and I. de Soysa, 2011, “Globalization and the empowerment of women: an analysis of spatial dependence via trade and foreign direct investment,” World Development, Vol. 39, No. 7, pp. 1065–1074.
- UNCTAD, 2014, Investment by TNCs and Gender: Preliminary Assessment and Way Forward,” Investment for Development Policy Research Series.
- World Bank, 2012, “World Development Report, Gender Equality and Development”. The World Bank Group, Washington D.C.

### Additional referenced works (selection)
- Anyanwu, J. C., 2016, “Analysis of Gender Equality in Youth Employment in Africa,” African Development Review, Vol. 28, No. 4, pp. 397–415.
- Björkman Nyqvist, M., and S. Jayachandran, 2017, "Mothers Care More, but Fathers Decide: Educating Parents about Child Health in Uganda," American Economic Review, 107(5): pp. 496–500.
- Castilla, C., and T. Walker, 2013, "Is Ignorance Bliss? The Effect of Asymmetric Information between Spouses on Intra-Household Allocations," American Economic Review, Vol. 103, No. 3, pp. 263–68.
- Chen, D. H. C., 2004, “Gender Equality and Economic Development: The Role for Information and Communication Technologies,” World Bank Policy Research Working Paper 3285, Washington DC.
- Chen, Z., Y. Ge, H. Lai, and C. Wan, 2013, “Globalization and Gender Wage Inequality in China,” World Development, Vol. 44, pp. 256–266.
- Crespo, N., and M. P. Fontoura, 2007, “Determinant Factors of FDI Spillovers – What Do We Really Know?”, World Development, Vol. 35, No. 3, pp. 410–425.
- Dieterich, C., Huang, A., and Thomas, A., 2016, Women’s opportunities and challenges in sub-Saharan African job markets, IMF Working Paper Series WP/16/118, Washington DC, USA.
- Juhn, C., G. Ujhelyi, and C. Villegas-Sanchez, 2013, “Trade Liberalization and Gender Inequality,” American Economic Review, Vol. 103, No. 3, pp. 269–273.
- Nelson, J., M. Porth, K. Valikai, and H. McGee, 2015, “A Path to Empowerment: The Role of Corporations in Supporting Women’s Economic Progress,” Corporate Social Responsibility Initiative at the Harvard Kennedy School and the U.S. Chamber of Commerce Foundation Corporate Citizenship Center.
- Stotsky, J. G., 2016, “Gender Budgeting: Fiscal Context and Current Outcomes,” International Monetary Fund (IMF), Working Paper No. 16/149.
- Stotsky, J.G., Shibuya, S., Kolovich, L., and Kebhaj, S., 2016, Trends in gender equality and women’s advancement, IMF Working Paper Series WP/16/21. Washington DC, USA

*Italicized attribution line appears in the source file.*

---

### Appendix II. List of Countries
- Afghanistan
- Albania
- Algeria
- Argentina
- Armenia
- Azerbaijan
- Bangladesh
- Belarus
- Belize
- Benin
- Bolivia
- Botswana
- Bulgaria
- Burundi
- Cambodia
- Cameroon
- Central African Republic
- Chad
- People's Republic of China
- Colombia
- Democratic Republic of the Congo
- Republic of the Congo
- Costa Rica
- Cote d'ivoire
- Dominican Republic
- Ecuador
- Egypt
- El Salvador
- Ethiopia
- Fiji
- The Gambia
- Georgia
- Ghana
- Guatemala
- Guinea
- Guyana
- Honduras
- Hungary
- India
- Indonesia
- Iran
- Iraq
- Jamaica
- Jordan
- Kazakhstan
- Kyrgyz Republic
- Lao people's Democratic Republic
- Lebanon
- Lesotho
- Liberia
- Libya
- Malawi
- Malaysia
- Maldives
- Mali
- Mauritania
- Mauritius
- Mongolia
- Morocco
- Mozambique
- Namibia
- Nepal
- Nicaragua
- Niger
- Pakistan
- Panama
- Paraguay
- Peru
- Philippines
- Romania
- Rwanda
- Senegal
- Serbia
- Sierra Leone
- South Africa
- Sri lanka
- Sudan
- Suriname
- Swaziland
- Syrian Arab Republic
- Republic of Macedonia
- Tajikistan
- Tanzania
- Thailand
- Togo
- Tonga
- Tunisia
- Turkey
- Uganda
- Ukraine
- United Kingdom (West Bank and Gaza listed separately in source)
- Venezuela
- Yemen

---

### Appendix III. Summary Statistics (Full sample)
- Log(FDI inflows) IMF STA database: Obs 1663; Mean 3.6; Std. Dev. 1.9; Min -7.3; Max 10.8
- Log(FDI inflows) IMF WEO database: Obs 1310; Mean 3.5; Std. Dev. 1.9; Min -5.5; Max 7.4
- Log(FDI inflows) UN database: Obs 1903; Mean 3.4; Std. Dev. 1.9; Min -4.6; Max 7.4
- GDI UNDP: Obs 2031; Mean 0.9; Std. Dev. 0.1; Min 0.5; Max 1.1
- GII UNDP: Obs 1844; Mean 0.5; Std. Dev. 0.1; Min 0.1; Max 0.8
- GDP growth rate IMF WEO database: Obs 1943; Mean 0.1; Std. Dev. 0.2; Min -2.3; Max 2.0
- Log(trade openness) World Bank: Obs 1967; Mean 4.2; Std. Dev. 0.7; Min -3.9; Max 5.8
- Polity 2 Polity IV Series: Obs 1938; Mean 2.6; Std. Dev. 6.0; Min -9.0; Max 10.0
- Log(public debt) IMF WEO database: Obs 1571; Mean 3.8; Std. Dev. 0.7; Min -2.4; Max 6.6
- Female population (in percentage of total population) World Bank: Obs 2031; Mean 50.3; Std. Dev. 1.0; Min 46.3; Max 53.8
- Government expenditures (in percentage of GDP) IMF WEO database: Obs 1772; Mean 26.6; Std. Dev. 9.2; Min 4.3; Max 105.3
- Natural resources rents (in percentage of GDP) World Bank: Obs 1986; Mean 8.9; Std. Dev. 11.3; Min 0.0; Max 66.5
- Rural population (in percentage of total population) World Bank: Obs 2031; Mean 54.2; Std. Dev. 20.1; Min 8.2; Max 94.6
- Female life expectancy World Bank Gender Statistics database: Obs 1985; Mean 66.8; Std. Dev. 10.2; Min 30.0; Max 82.1
- Average female year of schooling World Bank Gender Statistics database: Obs 1934; Mean 5.7; Std. Dev. 2.7; Min 0.8; Max 10.6
- Log(gross national income per capita) UNDP: Obs 2031; Mean 8.1; Std. Dev. 0.9; Min 6.0; Max 9.7
- Maternity mortality ratio (per 100,000 live births) World Bank Gender Statistics database: Obs 2031; Mean 322.4; Std. Dev. 363.0; Min 7.0; Max 2900.0
- Gross enrolment in secondary level (gender parity index) United Nations database: Obs 1323; Mean 0.9; Std. Dev. 0.2; Min 0.2; Max 1.6
- Female share of parliamentary seats World Bank Gender Statistics database: Obs 1553; Mean 15.1; Std. Dev. 10.3; Min 0.0; Max 63.8
- Female to male labor participation rate ratio International Labour Organization: Obs 1932; Mean 66.6; Std. Dev. 23.9; Min 14.0; Max 108.1
- Female to male ratio of employers (in percentage of employment) World Bank Gender Statistics database: Obs 655; Mean 0.4; Std. Dev. 0.2; Min 0.0; Max 1.9
- Female informal employment (in percentage of total employment) International Labour Organization: Obs 766; Mean 55.9; Std. Dev. 21.4; Min 4.2; Max 89.2
- Female to male ratio of part-time employment (in percentage of total employment) World Bank Gender Statistics database: Obs 811; Mean 2.1; Std. Dev. 0.9; Min 1.1; Max 5.3
- Gender wage gap International Labour Organization: Obs 962; Mean 16.6; Std. Dev. 14.3; Min -27.7; Max 47.2
- Gender budgeting dummy IMF Gender budgeting portal: Obs 2021; Mean 0.3; Std. Dev. 0.4; Min 0.0; Max 1.0
- Equal wage law dummy World Bank Gender Statistics database: Obs 1353; Mean 0.1; Std. Dev. 0.3; Min 0.0; Max 1.0
- Non-discrimination in hiring law dummy World Bank Gender Statistics database: Obs 1449; Mean 0.2; Std. Dev. 0.4; Min 0.0; Max 1.0
- Female access to resources OECD - Gender, Institutions and Development: Obs 1945; Mean 0.4; Std. Dev. 0.2; Min 0.0; Max 1.0
- Log(FDI inflows in percentage of GDP) IMF STA database: Obs 1227; Mean -3.9; Std. Dev. 1.6; Min -16.8; Max -0.2
- Log(FDI inflows in percentage of GDP) IMF WEO database: Obs 1095; Mean -3.5; Std. Dev. 0.8; Min -4.6; Max -0.2
- Log(FDI inflows in percentage of GDP) United Nations database: Obs 1907; Mean 0.7; Std. Dev. 1.3; Min -4.6; Max 4.0

---

### Appendix IV. Robustness Test 1 — Using WEO and United Nations Data on FDI (selected coefficients)
- Dependent variable GDI / GII (columns show alternating GDI and GII specifications):
  - Log(FDI inflows): 0.00314***; -0.00372***; 0.00275***; -0.00327***; 0.00289***; -0.00329***; 0.00269***; -0.00310*** (standard errors shown in table)
  - GDP growth: 0.00524; 0.0125; 0.00705; 0.0114; 0.00405; 0.00763; 0.00526; 0.00666
  - Log(trade openness): 0.00738**; 0.00920; 0.00851**; 0.00588; 0.0110***; 0.000977; 0.0102***; -0.00122
  - Polity2: 0.00130***; -0.000957*; 0.00107***; -0.000813; 0.000133; -0.000843**; 5.63E-06; -0.000890**
  - Log(public debt): -0.00649***; 0.00762***; -0.00742***; 0.00755***; -0.00614***; 0.00302; -0.00626***; 0.00346*
  - Female population: 0.00523***; -0.0145***; 0.00625***; -0.0125***
  - Government expenditures: 0.000733***; 0.000322; 0.000731***; 0.000189
  - Rural population: 0.000577**; -0.00150***; -0.000187; -0.00018
- Observations: 973; 973; 960; 960; 1,267; 1,267; 1,253; 1,253
- R-squared range shown: 0.963 to 0.973
- Country-fixed effects: YES; Year-fixed effects: YES
- Significance notation: Standard errors in parentheses. *** p<0.01; ** p<0.05; * p<0.1

---

### Appendix V. Robustness Test 2 — Alternative Econometric Method (selected coefficients)
- Dependent variables (GDI / GII):
  - Log(FDI inflows): 0.00606***; -0.00478**; 0.00570***; -0.00515**
  - GDP growth: 0.0042; 0.00867; 0.00479; 0.00706
  - Log(trade openness): 0.00560*; -0.0114**; 0.00534*; -0.0174***
  - Polity2: 0.000868***; 0.000146; 0.000732***; -0.000121
  - Log(public debt): -0.00324***; 0.00348*; -0.00338***; 0.00435**
  - Female population: -0.00124; -0.0105***
  - Government expenditures: 0.000632***; 0.000549**
  - Natural resources rents: -0.000114; 0.000785***
  - Observations: 1,111; 1,211; 1,100; 1,193
  - R-squared: 0.972; 0.960; 0.974; 0.962
  - Country-fixed effects: YES; Year-fixed effects: YES

---

### Appendix VI. Robustness Test 3 — Including Additional Control Variables (selected coefficients)
- Dependent variables (GDI / GII):
  - Log(FDI inflows): 0.00356***; -0.00218**; 0.00354***; -0.00214**
  - GDP growth: 0.00635; 0.0136; 0.00632; 0.0143
  - Log(trade openness): 0.0104***; -0.00394; 0.0105***; -0.00450
  - Polity2: 0.000881***; -0.000902**; 0.000878***; -0.000895*
  - Log(public debt): -0.00556***; 0.00569***; -0.00553***; 0.00559***
  - Female population: 0.00388**; -0.0110***; 0.00387**; -0.0109***
  - Government expenditure: 0.000658***; 0.000231; 0.000658***; 0.000236
  - Natural resources rents: -0.000231; 0.000875***; -0.000232; 0.000870***
  - Additional religion and ethnic controls with coefficients and significance shown in table (Ethnic diversity: -0.689***; 0.723***; Islam: -0.256***; 0.418***; Catholicism: 0.0421***; 0.0850***; Protestantism: 0.487**; 1.682***; Hinduism: -15.62***; -7.012; Buddhism: 0.0312; -0.0547)
  - Observations: 1,146; 1,146; 1,381; 1,381
  - R-squared: 0.970; 0.963; 0.970; 0.962
  - Country-fixed effects: YES; Year-fixed effects: YES

---

### Appendix VII. Robustness Test 4 — Using FDI in Percentage of GDP Instead of FDI Per Capita (selected coefficients)
- Dependent variables across many specifications (GDI / GII):
  - Log(FDI inflows): 0.00404***; -0.00350***; 0.00381***; -0.00310**; 0.00306***; -0.00473**; 0.00247**; -0.00486***; 0.00348***; -0.00335***; 0.00341***; -0.00328***
  - GDP growth coefficients listed across columns, e.g., 0.00805; 0.0110; 0.0103*; 0.00866; 0.00255; 0.0109; 0.00421; 0.00803; 0.00507; 0.00635; 0.00618; 0.00538
  - Log(trade openness) coefficients listed across columns, e.g., 0.00302; 0.0133**; 0.0040; 0.00999; 0.00527; 0.00912; 0.00497; 0.00569; 0.00978***; 0.0018; 0.00896***; -0.00034
  - Polity2: 0.00123***; -0.000845*; 0.000976***; -0.000721; 0.000792***; -0.000639; 0.000609**; -0.000339; 0.000124; -0.000840**; -3.52E-06; -0.000887**
  - Log(public debt): -0.00516***; 0.00701***; -0.00597***; 0.00691***; -0.00755***; 0.00683**; -0.00840***; 0.00635**; -0.00606***; 0.00314*; -0.00614***; 0.00355*
  - Female population and other controls vary across columns with given coefficients and significance
  - Observations: 899; 899; 890; 890; 882; 882; 872; 872; 1,269; 1,269; 1,255; 1,255
  - R-squared range reported: 0.962 to 0.975
  - Country-fixed effects: YES; Year-fixed effects: YES
  - Data source noted per column group: IMF's STA; IMF's WEO; United Nations

---

### Appendix VIII. Robustness Test 5 — Using Alternative Dependent Variables (selected coefficients)
- Dependent variables (GDI / GII alternative measures):
  - Log(FDI inflows): 0.0036***; -0.0024**; 0.0033***; -0.0020**
  - GDP growth: 0.000387; 0.0124; 0.0026; 0.0098
  - Log(trade openness): 0.0064**; -0.00090; 0.0092***; -0.0066
  - Polity2: 0.0008***; -0.0010**; 0.0007***; -0.0011**
  - Log(public debt): -0.0046***; 0.0065***; -0.0053***; 0.0076***
  - Female population: 0.0074***; -0.0155***
  - Government expenditure: 0.0006***; 0.0002
  - Natural resources rents: -0.0004**; 0.0008***
  - Observations: 1,039; 1,039; 1,029; 1,029
  - R-squared: 0.973; 0.963; 0.974; 0.964
  - Country-fixed effects: YES; Year-fixed effects: YES

---

*Italicized attribution line appears in the source file.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1825.pdf_