## wp1734

## Source details

**Canonical URL:** [wp1734](https://www.imf.org/-/media/files/publications/wp/wp1734.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/wp1734.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/wp1734.pdf.json)

---

### Abstract and contribution
- Women’s labor force participation (LFP) averages around 80 percent for men but only 50 percent for women globally.
- Latin America and the Caribbean (LAC) recorded the largest gains in female LFP in the world during 1990–2014.
- As women in LAC become more educated—and the share of women with a college education now exceeds that of men in several LAC economies—their productivity rises, increasing potential GDP losses if they remain outside the labor force.
- Paper’s three-fold contribution:
  - (i) Document current LAC female LFP levels and recent evolution.
  - (ii) Present “back-of-the-envelope” estimates for gains in GDP from higher female LFP.
  - (iii) Discuss policies shown to generate positive effects in female LFP.
- Focus and data:
  - Uses ILO estimates of LFP data from 1990-2014.
  - LFP defined as employed + unemployed aged 15+ divided by non-institutionalized population aged 15+.
  - LFP estimates include both formal and informal work.
  - Paper abstracts from the gender gap in wages due to data constraints.

### Key findings on trends in female and male LFP
- Female LFP:
  - 1990: 44 percent of women in LAC participated in the labor force.
  - 2014: 54 percent.
  - LAC improvement was largest across regions and broad-based across subregions, especially the five largest South American economies.
- Subregional heterogeneity:
  - Caribbean historically had high female LFP, above the United States.
  - Central America and the five largest South American economies started below 40 percent in the early 1990s; South America recorded significant gains, Central America caught up more modestly.
- Gender gap and male LFP:
  - Male LFP in LAC averages nearly 80 percent, versus 68 percent in advanced economies—this high male LFP is a key driver of the large gender gap.
- Regression/demographic exercise:
  - Cross-section regression of male LFP on log GDP per capita, (log GDP per capita)^2 and share of population over 65 using 2014 data for 170 countries.
  - Share of population over 65 is a significant predictor of male LFP.
  - Counterfactual: if each LAC country had the U.S. share of population over 65, gender gaps would be substantially reduced in most LAC subregions (except CAPDR).
  - For Brazil, Colombia, Peru, Uruguay demographics explain a significant part of LFP gaps vis-à-vis the U.S.; for Argentina, Chile, Mexico gender gaps remain significant.

### Development level and the U-shaped relationship
- U-shaped relationship between female LFP and log GDP per capita:
  - Low development: female LFP high due to agricultural work and income effects.
  - Middle-income: female LFP falls as substitution effect (childcare/home production) dominates.
  - High development: female LFP rises as women enter “brain-heavy” occupations and use professional childcare.
- Empirical patterns for LAC:
  - 1980: LAC countries tended to be below the estimated U-shaped curve.
  - 2010: majority of LAC countries moved above the estimated curve.
- Male LFP and development:
  - No U-shape; higher log GDP per capita linearly associated with lower male LFP.
  - Nearly all countries have male LFP between 60 and 80 percent (except a handful).

### Education, age, and female LFP
- Education drives female LFP by:
  - Increasing the probability that women work.
  - Raising productivity of those who work, reducing informal work.
- Women’s education in LAC rising; greater female education associated with higher female LFP.

### Education and labor force participation trends among younger women (LA-6 countries average)
- Data: household surveys for six largest LAC economies (survey years: 2013 for Chile and Brazil, 2014 for Mexico, 2015 for Argentina, Colombia and Peru); simple average of Brazil, Chile, Colombia, Mexico, Peru, and Uruguay; College = incomplete college or more.
- Education shares:
  - Women aged 25–34: 34 percent have a college education, compared to 31 percent of men aged 25–34.
  - Cohort aged 55–65: 15 percent of women had college educations, compared to 18 percent of men.
- LFP by age, sex, education (LA-6 average):
  - Higher educated (some college or more): almost no gender difference in LFP in the 15–24 age group; gender gap increases with age.
  - High school or less: persistent gender gap across ages.

### Policies that could raise female LFP — approach and limitations
- Short-term policy levers identified: provision of childcare, maternity/paternity leave.
- Empirical strategy:
  - Cross-section of countries in 2014 to examine associations between policies and gender gaps in LFP.
  - Panel regressions 1990–2010 using women’s legal rights (only policy variable with sufficient historical data).
- Key caveats:
  - Cross-section associations are not causal and may suffer omitted variable bias and endogeneity.
  - Scarcity of household survey data in LAC and limited time series for policy variables constrain causal analysis.

### Summary statistics: LAC versus advanced economies (selected statistics)
- LFP gap:
  - LAC (22 countries) mean: 26.29; Std. Dev.: 7.86; Min: 10; Max: 39.8
  - Advanced (29 countries) mean: 13.56; Std. Dev.: 4.01; Min: 7.1; Max: 22
- Log GDP per capita:
  - LAC mean: 8.31; Std. Dev.: 0.81; Min: 6.2; Max: 9.57
  - Advanced mean: 10.36; Std. Dev.: 0.5; Min: 9.18; Max: 11.33
- Population ages 65 and above:
  - LAC mean: 7.41; Std. Dev.: 2.4; Min: 4.58; Max: 14.33
  - Advanced mean: 16.99; Std. Dev.: 3.51; Min: 11.03; Max: 25.71
- Roads quality (WEF):
  - LAC mean: 3.54; Std. Dev.: 0.85; Min: 2.25; Max: 5.23
  - Advanced mean: 5.21; Std. Dev.: 0.77; Min: 3.31; Max: 6.22
- Female years of schooling (Barro-Lee):
  - LAC mean: 7.67; Std. Dev.: 1.82; Min: 3.08; Max: 10.33
  - Advanced mean: 11.29; Std. Dev.: 1.16; Min: 9; Max: 13.03
- Log of paid days of maternity leave (World Bank):
  - LAC mean: 4.49; Std. Dev.: 0.3; Min: 3.74; Max: 5.2
  - Advanced mean: 4.52; Std. Dev.: 0.99; Min: 0; Max: 5.47
- Are parents entitled to flexible/part-time schedules? (WB):
  - LAC mean: 0; Std. Dev.: 0; Min: 0; Max: 0
  - Advanced mean: 0.45; Std. Dev.: 0.51; Min: 0; Max: 1
- Does the law mandate equal remuneration for work of equal value? (WB):
  - LAC mean: 0.29; Std. Dev.: 0.46; Min: 0; Max: 1
  - Advanced mean: 0.69; Std. Dev.: 0.47; Min: 0; Max: 1
- Does the law mandate nondiscrimination based on gender in hiring? (WB):
  - LAC mean: 0.33; Std. Dev.: 0.48; Min: 0; Max: 1
  - Advanced mean: 0.69; Std. Dev.: 0.47; Min: 0; Max: 1
- Are childcare payments tax deductible? (WB):
  - LAC mean: 0.14; Std. Dev.: 0.36; Min: 0; Max: 1
  - Advanced mean: 0.34; Std. Dev.: 0.48; Min: 0; Max: 1
- Does the government support or provide childcare services? (WB):
  - LAC mean: 0.95; Std. Dev.: 0.22; Min: 0; Max: 1
  - Advanced mean: 1; Std. Dev.: 0; Min: 1; Max: 1
- Log of paid days of paternity leave (WB):
  - LAC mean: 1.06; Std. Dev.: 0.88; Min: 0; Max: 2.64
  - Advanced mean: 1.27; Std. Dev.: 1.34; Min: 0; Max: 4.5

### Cross-section regression findings (gender gap in LFP)
- Log GDP per capita and its square are the strongest predictors of the gender gap in LFP:
  - Column (1): Log GDP pc coefficient 61.11*** (t = 6.35); Log GDP pc square -3.623*** (t = -6.46); R-squared 0.308; Observations 77.
  - GDP differences explain about a third of variation in the gender gap in LFP (R^2 ≈ 0.308).
- Population 65+ (%) is not significant in cross-section specifications.
- Women’s legal rights index and female years of schooling:
  - Legal rights index negative (lower gap) in various specifications; e.g., column (6) legal rights index = -2.689* (t = -2.03).
  - Years of schooling (women) negative but not strongly significant (e.g., -1.704, t-statistics around -1.69).
- Childcare tax deductibility, log maternity days, log paternity days show mixed, non-robust relationships:
  - Log maternity days positive in some specifications (e.g., 1.697, t ≈ 1.90) — interpretation cautioned for endogeneity.
  - Log paternity days negative in one specification (-1.618, t ≈ -1.09) but not significant.

### Panel regression findings (1990–2010) — women’s legal rights
- Legal rights index robustly predicts lower gender gaps in LFP in panel regressions:
  - Column (1): Log GDP pc 15.91*** (t = 16.32); Log GDP pc square -0.922*** (t = -14.50); Legal rights index -0.392*** (t = -5.62); Observations 4,294; R-squared 0.496.
- Significant negative subcomponents:
  - Sons and daughters equal inheritance: -2.834*** (t = -5.53).
  - No joint titling (not having joint titling as default): -1.999*** (t = -4.07).
  - Guaranteed gender equality by Constitution: -2.637*** (t = -6.10).
- Population 65+ (%) negative and significant in some panel specifications (e.g., -0.214*, -0.229*).

### Interpretation and mechanisms
- Associations:
  - Women’s legal rights and female education associated with lower gender gaps in LFP in cross-section and panel evidence.
  - Causality not definitively established from cross-section results.
- Proposed mechanisms:
  - Legal rights (asset control, constitutional protection) may increase women’s incentives and ability to participate in market work.
  - Education increases human capital and returns in the labor market, raising female LFP and reducing gender gap.
- Cautions on leave and childcare policy estimates:
  - Positive association between longer maternity leave and the gender gap may reflect endogeneity, limited coverage, or employer hiring responses.
  - Cross-country regressions susceptible to omitted variable bias and reverse causality; household-level studies provide stronger identification.

### Evidence from national/household-level studies
- Country-level studies evaluate leave policies, childcare subsidies, preschool programs.
- Policy interactions matter: marginal effect of expanding one policy depends on generosity and presence of other family-related benefits.
- Childcare subsidies and vouchers generally successful in supporting higher female LFP or increasing women’s hours worked.

### Childcare and leave policies: evidence and effects
- Childcare provision: “consistently shown as one of the most effective policies” to incentivize women entering the labor force.
- More generous childcare policies increase female LFP in advanced economies; high-quality childcare helps mothers return to previous jobs or increase part-time hours.
- Leave policies:
  - Two types studied: changing duration of paid leave; changing income replacement rates with duration fixed.
  - Longer maternity leave tends to be associated with higher women’s employment (Ruhm, 1998; Jaumotte, 2004).
  - Extending job-protected maternity leave from one to two years associated with delayed return to work and potential fertility increases without significant long-run earnings consequences (Lalive and Zweimuller, 2009).
  - Increasing replacement rates had negligible effect on mothers’ job continuity in Japan (Asai, 2015), possibly due to insufficient childcare options.
- Table 4 examples (Mateo Diaz and Rodriguez Chamussy, 2013):
  - Guatemala — Hogares Comunitarios — LFP↑: No — Hours worked↑: Yes
  - Mexico — Estancias Infantiles para madres trabajadoras — LFP↑: Yes — Hours worked↑: Yes
  - Colombia — Hogares Comunitarios de Bienestar Familiar — LFP↑: Yes — Hours worked↑: Yes
  - Ecuador — Centros de desarollo infantil — LFP↑: Yes — Hours worked↑: (blank)
  - Chile — Chile crece contigo — LFP↑: Unclear — Hours worked↑: (blank)
  - Brazil — Rio de Janeiro’s public daycare program — LFP↑: Yes — Hours worked↑: No

### Other policies with potential to raise female labor supply
- Greater flexibility in work hours including temporary part-time employment.
- Removing the second-earner penalty from family taxation.
- “In-work” tax credits for low wage female workers.
- Gender budgeting to free up and allocate fiscal resources toward gender issues (examples documented for Mexico, Ecuador, Bolivia, El Salvador).

### Macroeconomic impact of increasing female LFP — methodology and headline estimates
- Regression setup:
  - Period: 1990—2014.
  - Sample: unbalanced panel of 169 countries (including 28 LAC countries).
  - Dependent variable: log of GDP per capita.
  - Regressors: female LFP, fertility, inflation, FDI, trade openness.
- Two counterfactuals:
  - Raise female LFP to Nordic average: 61 percent (Denmark, Finland, Iceland, Norway, Sweden).
  - Raise female LFP to equal country’s male LFP (gender parity).
- Key findings:
  - Gain in GDP per capita from raising female LFP to Nordic average could be as high as 10 percent (Suriname) in specific cases.
  - Gender parity exercise can yield gains as large as 20 percent in some countries.
  - For LAC on average, estimates suggest gains between 4 and 14 percent of GDP if they reach the Nordic average or achieve gender parity, respectively.
  - Caveat: assumption that male LFP remains unchanged is likely unrealistic; male LFP could decline, dampening gains.
- Comparisons:
  - Jacobsen (2013): ~7 percent worldwide from higher female LFP.
  - Aguirre: 12 percent for Argentina, 9 percent for Brazil.
  - Woetzel et al (2015): 14 percent for Latin America.
  - Some LAC countries already exceed Nordic female LFP (St. Lucia, Bolivia, Barbados, Peru); assumed no gain in those cases.

### Female LFP during economic downturns
- Theory: women often act as added workers when primary earners lose employment → counter-cyclical female LFP.
- Empirical findings:
  - Added worker effects stronger among poorer, lower-educated women and those married to men with lower income/education.
  - Country examples of added worker effects: Mexico (mid-1990s), Peru (early 1980s), Argentina (early 1990s).
- Aggregate empirical test:
  - Fixed effects regressions of female (or male) LFP on a significant slowdown dummy and deviation of log GDP per capita from HP-filtered trend; year fixed effects included.
  - Significant slowdown dummy = 1 when deviation from HP trend of log GDP per capita is in bottom 10% of sample (1990–2014).
  - Results (Figure 13):
    - Latin America: evidence consistent with higher female LFP during significant downturns; men in LAC also more present during downturns.
    - EM Asia: small, not statistically significant effects.
    - Advanced economies: responses pro-cyclical (weaker added worker effects) consistent with higher savings and stronger safety nets.
  - Authors caution results are suggestive and call for detailed household-level data.

### Conclusions and policy recommendations
- Summary:
  - Women’s work is an important untapped resource globally and in LAC.
  - LAC increased female LFP from under 45 percent in 1990 to levels similar to the U.S. and EM Asia by 2014; gains driven by South American economies; Central America trails.
  - Contributing factors include economic development, higher female education, and legal rights.
  - Gender gap in LFP in LAC remains sizable, partly because male LFP is relatively high due to young demographics.
  - Younger, more educated women are working more and closing the gender LFP gap.
- Policy recommendations:
  - Strengthen women’s legal rights, especially property rights and anti-discrimination laws.
  - Implement/extend early childhood education and childcare services.
  - Promote girls’ education and job training to boost higher education attainment.
  - Improve infrastructure, especially urban transportation.
  - Guarantee equal pay for similar work and strengthen women’s household decision-making power.
- Expected broader benefits:
  - Policies supporting working women increase women’s bargaining power; women tend to allocate household resources toward children’s education and healthcare, raising labor productivity and boosting the economy.

### Annex highlights
- Panel regression results for growth (dependent variable: Log GDP pc), selected coefficients:
  - FLFP: 0.00523*** (4.48)
  - Fertility: -0.228*** (-24.94)
  - Inflation: -0.00000579 (-0.96)
  - FDI: -0.000379 (-0.86)
  - Trade openness: 0.00108*** (5.53)
  - Constant: 8.408*** (109.98)
  - Observations: 3721
  - R-squared: 0.202
  - Significance: * p<0.05  ** p<0.01  *** p<0.001
- Data sources include WDI (GDP per capita constant 2005 US$, female/male LFP, fertility, inflation, FDI, trade openness), Barro-Lee (female years of schooling), World Economic Forum (roads quality), Women, Business and the Law (legal rights index, maternity/paternity leave, childcare tax deductibility).

*IMF Working Paper WP/17/34 — Prepared by Natalija Novta and Joyce Cheng Wong, February 2017*

### Section 1

### Women at Work in Latin America and the Caribbean — Section 1

### Abstract and contribution
- Women’s labor force participation (LFP) averages around 80 percent for men but only 50 percent for women globally, implying nearly half of women’s productive potential remains untapped compared to one-fifth for men.
- Latin America and the Caribbean (LAC) saw the largest gains in female LFP in the world during the last two decades.
- As women in LAC become more educated—and the share of women with a college education now exceeds that of men in several LAC economies—their productivity at work becomes higher, increasing potential GDP losses if they remain outside the labor force.
- Paper’s three-fold contribution:
  - (i) Document where LAC is today in terms of female LFP and its recent evolution.
  - (ii) Present simple “back-of-the-envelope” estimates for gains in GDP from higher female LFP.
  - (iii) Discuss policies shown to generate positive effects in female LFP.
- Focus and data:
  - Uses ILO estimates of LFP data from 1990-2014.
  - LFP defined as sum of employed and unemployed aged 15 and above, divided by total non-institutionalized population aged 15 and above.
  - LFP estimates do not distinguish formal or informal work (both included).
  - Due to data constraints, the paper abstracts from the gender gap in wages.

### Key findings on trends in female and male LFP
- In 1990 only 44 percent of women in LAC participated in the labor force.
- In 2014 this ratio increased to 54 percent, close to levels seen in the United States and EM Asia.
- The improvement in LAC was the largest of any region in the world and was driven by a broad-based improvement in all LAC subregions, especially the five largest South American economies.
- Heterogeneity across subregions:
  - Caribbean countries historically had high female LFP rates, even above those of the United States.
  - Central America and the five largest South American economies started with female LFP below 40 percent in the early 1990s; South America recorded significant gains, Central America caught up at a more modest rate.
- The gender gap in LFP (male minus female LFP) remains large in LAC versus other regions despite improvements.
  - Male LFP in LAC averages nearly 80 percent, versus 68 percent in advanced economies — this very high male LFP is a key driver of the large gender gap.
- Regression and demographic exercise:
  - Cross-section regression of male LFP on log GDP per capita, the square of log GDP per capita and the share of population over 65, using 2014 data for 170 countries.
  - The share of population over 65 is a significant predictor of male LFP.
  - A counterfactual assuming each LAC country has the same share of population over 65 as the U.S. implies substantially reduced gender gaps in most LAC subregions (except CAPDR).
  - For some large LAC economies (Brazil, Colombia, Peru, Uruguay) demographics account for a significant part of their LFP gaps vis-à-vis the U.S.; for others (Argentina, Chile, Mexico) gender gaps remain significant.

### Development level and the U-shaped relationship
- There is a well-documented U-shaped relationship between female LFP and log GDP per capita.
  - At low development levels, female LFP is high due to agricultural work and income effects.
  - At middle-income levels, female LFP falls as the substitution effect (childcare and home production) dominates.
  - At higher development and education levels, female LFP rises as women enter “brain-heavy” occupations and use professional childcare.
- Empirical patterns for LAC:
  - In 1980, LAC countries tended to be below the estimated U-shaped curve (female LFP lower than predicted by development level).
  - By 2010 a majority of LAC countries moved above the estimated curve, indicating improvements beyond those explained by development alone.
- Male LFP and development:
  - The relationship between log GDP per capita and male LFP exhibits no U-shape; instead, higher log GDP per capita is linearly associated with lower male LFP.
  - Variation in male LFP is relatively small: nearly all countries (except a handful) have male LFP rates between 60 and 80 percent.
  - Interpretation: as men’s wages increase with GDP per capita, the substitution effect appears to lead men to shift into leisure.

### Education, age, and female LFP
- Education is a significant driver of female LFP and plays a dual role:
  - Drives more women to work.
  - Increases the productivity of those who work, reducing likelihood of informal unprotected work.
- Women’s education levels in Latin America are on the rise, and greater female education tends to be associated with higher female LFP.

*IMF Working Paper WP/17/34 — Prepared by Natalija Novta and Joyce Cheng Wong, February 2017*

### Section 2

### wp1734 - Section 2

### Education and labor force participation trends among younger women
- Using latest available household survey data for the six largest LAC economies (survey years included 2013 (Chile and Brazil), 2014 (Mexico) and 2015 (Argentina, Colombia and Peru)):
  - Women aged 25–34: on average, 34 percent have a college education, compared to 31 percent of men in the same age group.
  - Cohort aged 55–65: 15 percent of women had college educations, compared to 18 percent of men.
  - Note: Simple average of Brazil, Chile, Colombia, Mexico, Peru, and Uruguay; College = incomplete college or more.
- Labor force participation (LFP) by age, sex, and education (simple average of Brazil, Chile, Colombia, Mexico, Peru, and Uruguay):
  - For those with some college education or more (higher educated), almost no gender difference in LFP in the 15–24 age group; the gender gap increases broadly with age.
  - For those with high school education or less, a more persistent gender gap exists throughout ages.
  - Note: High school or less = high school education or less.

### Policies that could raise female LFP — approach and limitations
- Short-term policy levers identified in literature: provision of childcare, maternity/paternity leave.
- Empirical strategy used:
  - Cross-section of countries in 2014 (advanced economies and emerging markets) to examine associations between policies and gender gaps in LFP.
  - Panel regressions for 1990-2010 using women’s legal rights (only policy variable with sufficient historical data).
- Key caveats:
  - Cross-section associations cannot be interpreted as causal and are likely subject to omitted variable bias and endogeneity.
  - Household survey data scarcity in LAC and lack of long time series for many policy variables limit causal analysis.

### Summary statistics: LAC versus advanced economies (Table 1 highlights)
- LFP gap:
  - LAC (22 countries) mean: 26.29; Std. Dev.: 7.86; Min: 10; Max: 39.8
  - Advanced (29 countries) mean: 13.56; Std. Dev.: 4.01; Min: 7.1; Max: 22
- Log GDP per capita:
  - LAC mean: 8.31; Std. Dev.: 0.81; Min: 6.2; Max: 9.57
  - Advanced mean: 10.36; Std. Dev.: 0.5; Min: 9.18; Max: 11.33
- Population ages 65 and above:
  - LAC mean: 7.41; Std. Dev.: 2.4; Min: 4.58; Max: 14.33
  - Advanced mean: 16.99; Std. Dev.: 3.51; Min: 11.03; Max: 25.71
- Roads quality (WEF):
  - LAC mean: 3.54; Std. Dev.: 0.85; Min: 2.25; Max: 5.23
  - Advanced mean: 5.21; Std. Dev.: 0.77; Min: 3.31; Max: 6.22
- Female years of schooling (Barro-Lee):
  - LAC mean: 7.67; Std. Dev.: 1.82; Min: 3.08; Max: 10.33
  - Advanced mean: 11.29; Std. Dev.: 1.16; Min: 9; Max: 13.03
- Log of paid days of maternity leave (World Bank):
  - LAC mean: 4.49; Std. Dev.: 0.3; Min: 3.74; Max: 5.2
  - Advanced mean: 4.52; Std. Dev.: 0.99; Min: 0; Max: 5.47
- Are parents entitled to flexible/part-time schedules? (WB):
  - LAC mean: 0; Std. Dev.: 0; Min: 0; Max: 0
  - Advanced mean: 0.45; Std. Dev.: 0.51; Min: 0; Max: 1
- Does the law mandate equal remuneration for work of equal value? (WB):
  - LAC mean: 0.29; Std. Dev.: 0.46; Min: 0; Max: 1
  - Advanced mean: 0.69; Std. Dev.: 0.47; Min: 0; Max: 1
- Does the law mandate nondiscrimination based on gender in hiring? (WB):
  - LAC mean: 0.33; Std. Dev.: 0.48; Min: 0; Max: 1
  - Advanced mean: 0.69; Std. Dev.: 0.47; Min: 0; Max: 1
- Are childcare payments tax deductible? (WB):
  - LAC mean: 0.14; Std. Dev.: 0.36; Min: 0; Max: 1
  - Advanced mean: 0.34; Std. Dev.: 0.48; Min: 0; Max: 1
- Does the government support or provide childcare services? (WB):
  - LAC mean: 0.95; Std. Dev.: 0.22; Min: 0; Max: 1
  - Advanced mean: 1; Std. Dev.: 0; Min: 1; Max: 1
- Log of paid days of paternity leave (WB):
  - LAC mean: 1.06; Std. Dev.: 0.88; Min: 0; Max: 2.64
  - Advanced mean: 1.27; Std. Dev.: 1.34; Min: 0; Max: 4.5
- Note: All variables marked (WB) come from Women, Business and the Law, World Bank (2016).

### Cross-section regression findings (Table 2)
- Log GDP per capita and its square are the most significant predictors of the gender gap in LFP:
  - Column (1): Log GDP pc coefficient 61.11*** (t = 6.35); Log GDP pc square -3.623*** (t = -6.46); R-squared 0.308; Observations 77.
  - Across specifications, coefficients remain highly significant (*** p<0.001).
- GDP differences alone explain about a third of variation in the gender gap in LFP (R^2 in column 1 ≈ 0.308).
- Population 65+ (%) is not a statistically significant predictor of the gender gap in LFP in cross-section specifications.
- Roads quality coefficients are positive but not statistically significant in the cross-section.
- Women’s legal rights index and female years of schooling:
  - Legal rights index coefficients are negative in various specifications; in column (6) legal rights index = -2.689* (t = -2.03).
  - Years of schooling (women) coefficients negative but not strongly significant (e.g., -1.704, t-statistics around -1.69).
- Childcare tax deductibility, log of maternity days, and log of paternity days show mixed and non-robust relationships:
  - Log of maternity days in some specifications is positive (e.g., 1.697, t ≈ 1.90) — interpretation cautioned due to endogeneity.
  - Log of paternity days negative in one specification (-1.618, t ≈ -1.09) but not significant.

### Panel regression findings (1990-2010) — women’s legal rights
- Legal rights index is a robust predictor of lower gender gaps in LFP in panel regressions:
  - Column (1): Log GDP pc 15.91*** (t = 16.32); Log GDP pc square -0.922*** (t = -14.50); Legal rights index -0.392*** (t = -5.62); Population 65+ (%) not significant in column (1); Observations 4,294; R-squared 0.496.
- Sub-components of the legal rights index with significant negative associations:
  - Sons and daughters equal inheritance: -2.834*** (t = -5.53).
  - No joint titling (i.e., not having joint titling as default in married couples): -1.999*** (t = -4.07).
  - Guaranteed gender equality by Constitution: -2.637*** (t = -6.10).
- Population 65+ (%) is negative and significant in some panel specifications (e.g., -0.214*, -0.229*).

### Interpretation and mechanisms
- Women’s legal rights and female education are associated with lower gender gaps in LFP in cross-section and panel evidence, though causality cannot be definitively established from cross-section results.
- Possible mechanisms:
  - Legal rights (asset control, constitutional protection) may increase women’s incentives and ability to participate in market work.
  - Education allows women to develop human capital and increases returns in the labor market, which tends to raise female LFP and reduce the gender gap.
- Cautions regarding leave and childcare policy estimates:
  - Positive association between length of maternity leave and the gender gap in cross-section may reflect endogeneity, selection effects, limited coverage of benefits, or employer hiring responses.
  - Cross-country regressions are prone to omitted variable bias and reverse causality; national and household-level studies provide more credible identification.

### Evidence from national/household-level studies (literature overview)
- Country-level studies generally evaluate three policy types: leave policies, childcare subsidies, preschool programs.
- Interactions across policies matter: the marginal effect of expanding one policy (e.g., maternity leave) depends on the generosity and presence of other family-related benefits (Cascio et al, 2015).
- Childcare subsidies and vouchers:
  - Generally successful in supporting higher female LFP or increasing women’s hours worked.

*Source: wp1734 - Section 2 (pdf: wp1734 - Section 2).*

### Section 3

### wp1734 - Section 3

### Childcare and leave policies: evidence and effects
- Childcare provision has been “consistently shown as one of the most effective policies” to incentivize women entering the labor force (Mateo Diaz and Rodriguez Chamussy, 2013).  
- More generous childcare policies have been shown to increase female labor force participation in advanced economies.  
- High-quality childcare tends to free mothers to return to previous jobs or increase hours in part-time work.  
- Leave policies increase women’s attachment to the labor force, but results are more context-dependent than childcare policies:
  - Two types of leave policy changes studied: (i) changing duration of paid leave; (ii) changing income replacement rates while holding duration fixed.
  - Longer maternity leave tends to be associated with higher women’s employment (Ruhm, 1998; Jaumotte, 2004).
  - Extending job-protected maternity leave from one to two years is associated with a delay in women’s return to work and potential fertility increases, without significant long-run earnings consequences (Lalive and Zweimuller, 2009).
  - Increasing replacement rates had only negligible effect on mothers’ job continuity rates in Japan (Asai, 2015), possibly due to insufficient childcare options.

- Table 4 — Childcare programs in LAC and their effects on female LFP and hours worked (Source: Mateo Diaz and Rodriguez Chamussy, 2013):
  - Guatemala — Hogares Comunitarios — LFP↑: No — Hours worked↑: Yes
  - Mexico — Estancias Infantiles para madres trabajadoras — LFP↑: Yes — Hours worked↑: Yes
  - Colombia — Hogares Comunitarios de Bienestar Familiar — LFP↑: Yes — Hours worked↑: Yes
  - Ecuador — Centros de desarollo infantil — LFP↑: Yes — Hours worked↑: (blank)
  - Chile — Chile crece contigo — LFP↑: Unclear — Hours worked↑: (blank)
  - Brazil — Rio de Janeiro’s public daycare program — LFP↑: Yes — Hours worked↑: No

### Other policies with potential to raise female labor supply
- Policies with less evidence but potential effects, especially for working mothers who respond to labor market or tax incentives:
  - Greater flexibility in work hours including temporary part-time employment (discussion in Jaumotte, 2004).
  - Removing the second-earner penalty from family taxation (Stotsky, 1997).
  - “In-work” tax credits for low wage female workers (IMF, 2013).
  - Sensitivity to gender budgeting issues (examples documented in Fragoso and Enriquez 2016 for Mexico, Ecuador, Bolivia, and El Salvador) to free up and ensure allocation of fiscal resources towards gender issues.

### Macroeconomic impact of increasing female LFP — methodology and headline estimates
- The paper provides a “back-of-the-envelope” estimate of the potential macroeconomic impact of increasing female LFP, based on fixed effects regressions and holding other variables constant.
- Regression setup:
  - Period: 1990—2014.
  - Sample: unbalanced panel of 169 countries, including 34 advanced countries, 28 LAC countries, 22 EM Asia countries, 21 MENAP countries, 40 Sub-Saharan African countries, 11 EM Europe.
  - Dependent variable: log of GDP per capita.
  - Regressors: (i) female LFP, (ii) fertility, (iii) inflation, (iv) FDI, (v) trade openness.
  - Two benchmark counterfactuals for female LFP increases:
    - Raise each country’s female LFP to the average female LFP of Nordic countries (61 percent; Denmark, Finland, Iceland, Norway, Sweden).
    - Raise each country’s female LFP to equal the country’s own male LFP (gender parity in LFP).

- Key findings from the exercise:
  - Gain in GDP per capita from increasing female LFP to the Nordic average could be as high as 10 percent (in Suriname), but generally depends on country and existing female LFP levels.
  - In the gender parity exercise (female LFP → male LFP), gains in GDP per capita are larger (as large as 20 percent in some cases).
  - Caveat: assuming female LFP reaches current male levels with male LFP unchanged is likely unrealistic—male LFP could decline in response, dampening GDP gains; this is not captured in the simple regressions.
  - On average for LAC countries, estimates suggest gains between 4 and 14 percent of GDP if they reach the Nordic average for female LFP, or if gender parity is achieved, respectively.
  - These “back-of-the-envelope” estimates are broadly similar to model-based estimates in Cuberes and Teignier (2016), which also assume gender parity and closing other gender gaps.

- Comparisons with other studies:
  - Jacobsen (2013): higher female LFP would raise GDP per capita by about 7 percent worldwide.
  - Aguirre: gains of 12 and 9 percent for Argentina and Brazil, respectively.
  - Woetzel et al (2015): estimate a gain of 14 percent for Latin America.
  - Note: some LAC countries already have female LFP higher than the Nordic average (St. Lucia, Bolivia, Barbados, Peru); in these cases, the authors assumed no GDP per capita gain.

### Female LFP during economic downturns
- Theory and evidence:
  - Women are often the second earner and may provide insurance to household income when the primary earner is unemployed → counter-cyclical (added worker) response of female LFP.
  - Empirical findings show women’s LFP can respond counter-cyclically to shocks, especially among poorer or lower-educated women, and those married to men with lower income/education (Bhalotra and Umana-Aponte, 2009; Sabarwal et al, 2010).
  - Economic downturns may negatively affect child schooling and survival in low-income countries with weak safety nets; cash transfers can smooth family consumption during downturns.

- Country examples of added worker effects during past crises in LAC:
  - Mexico during mid-1990s (Parker and Skoufias, 2006; Cunningham, 2001; McKenzie, 2003).
  - Peru in early 1980s (Francke, 1992).
  - Argentina in early 1990s (Cerutti, 2000).
  - Preliminary evidence for 2009 global recession impacts on female LFP in Latin America is suggested but requires country-level analysis (Cho and Newhouse 2013).

- Aggregate empirical test in the paper:
  - Fixed effects regressions of female (or male) LFP on: (i) a dummy for significant slowdown; (ii) deviation of log GDP per capita from its HP-filtered trend.
  - Significant slowdown dummy equals 1 when deviation from HP trend of log GDP per capita is in the bottom 10% of the entire sample (1990–2014, countries in LAC, Asia, Europe).
  - Regressions include year fixed effects; separate regressions run for Latin America, advanced economies, and EM Asia.
  - Results (Figure 13):
    - For Latin America, results are consistent with higher LFP during significant downturns, especially strong for women (men in LAC also more present during downturns).
    - Potential added worker effects are most pronounced in Latin America; effects in Asia are small and not statistically significant.
    - In advanced economies, responses are pro-cyclical (weaker added worker effects), consistent with higher savings and stronger safety nets.
  - Authors caution that these results are suggestive and that more detailed household-level data are needed to investigate mechanisms.

### Conclusions and policy recommendations
- Summary assessments:
  - Women’s work is an important untapped resource globally and in LAC; increasing female LFP can help boost income and employment.
  - LAC increased female LFP from under 45 percent in 1990 to broadly US and emerging Asia levels by 2014, with gains driven by South American economies; Central America trails behind.
  - Contributing factors: economic development, higher female educational attainment, and equal legal rights.
  - Gender gap in LFP within LAC remains sizable, partly because men in LAC participate more due to relatively young demographics.
  - Younger and more educated women are working more and closing the gender LFP gap; female LFP gains are gradual but the region is on the right track.

- Policy recommendations (based on empirical work and literature):
  - Where needed, strengthen women’s legal rights, especially property rights and laws protecting women from discrimination.
  - Implement/extend early childhood education and childcare services.
  - Promote girls’ education and implement educational and job training policies to further women’s higher education.
  - Improve infrastructure, especially urban transportation.
  - Additional supportive measures (mentioned earlier in the section) include policies guaranteeing equal pay for similar work and other measures that strengthen women’s household decision-making power.

- Expected broader benefits:
  - Policies supporting working women increase women’s bargaining power in households; women tend to allocate household resources toward children’s education and healthcare, thereby benefiting families, increasing labor productivity, and boosting the economy.

*Source: wp1734 - Section 3 (selected pages).*

### Section 4

### wp1734 - Section 4

### Annex I: Labor Force Participation by education, gender and age (LA-6)
- Annex I is titled "Labor Force Participation by education, gender and age (LA-6)". (No tabular content present in this excerpt.)

### Annex II: Panel Regression Results for Growth
- Dependent variable: Log GDP pc
- Estimated coefficients (t statistics in parentheses):
  - FLFP: 0.00523*** (4.48)
  - Fertility: -0.228*** (-24.94)
  - Inflation: -0.00000579 (-0.96)
  - FDI: -0.000379 (-0.86)
  - Trade openness: 0.00108*** (5.53)
  - Constant: 8.408*** (109.98)
- Sample and fit:
  - Observations: 3721
  - R-squared: 0.202
- Significance notation:
  - * p<0.05  ** p<0.01  *** p<0.001

### Annex III: Data Sources
- 1) Output (WDI): GDP per capita, constant 2005 US$
- 2) Famale/male labor force participation (WDI): Female/Male labor force participation (% of female/male population ages 15+)
- 3) Education (Barro-Lee): female years of schooling
- 4) Population (WDI)
- 5) Roads quality (World Economic Forum)
- 6) Legal rights index (Women, Business and the Law, 2012): subcomponents include 1) unmarried equal property rights for immoveables,2) married equal property rights for immoveables, 3) sons and daughters equal inheritance for immoveables, 4) surviving spouse’s equal inheritance for immoveables, 5) no joint titling as default in married couples, 6) adult married women heal of household or head of family, 7) married women can get a job or pursue profession, 8) married women can open bank account, 9) married women can sign contract, 10) married women can initiate legal proceedings without housband’s permission, 11) guarnateed gender equality by Consitution.
- 7) Childcare tax deductible (Women, Business and the Law, 2016)
- 8) Maternity leave days (Women, Business and the Law, 2016)
- 9) Paternity leave days (Women, Business and the Law, 2016)
- 10) Fertility (WDI): total births per woman
- 11) Inflation (WDI): consumer prices, annual %
- 12) Foreign Direct Investment (WDI): net inflows, % of GDP
- 13) Trade openness (WDI): Exports + imports of goods and services, % of GDP

*Source: wp1734 - Section 4 (excerpt).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/wp1734.pdf_