The Other Gender Gap

In 152 of 203 countries, more women than men are enrolled in higher education. The median GPI is 1.2, but the population-weighted mean is only 1.1—the giants lag behind.

Distribution

Three-quarters of countries (75%) now have more women than men in tertiary education.

By Region

Latin America leads (pop-weighted GPI 1.3); Sub-Saharan Africa lags (0.8).

By Income Group

The income gradient is monotonic: low-income 0.7 → high-income 1.2.

Large Countries

Japan and South Korea are outliers among rich countries—still male-majority.

Historical Trends

Using a balanced panel of 30 countries with data from 1990 onward, we track the share with GPI > 1 over time. The trend doubled: from 33% in 1990 to 67% by 2024. A LOWESS trend line shows the overall trajectory.

Education vs Employment

In several MENA countries, women outnumber men in universities while overall female workforce participation remains low.

Data

Sources:

Indicator	World Bank Code	Description
Tertiary GPI	SE.ENR.TERT.FM.ZS	Ratio of female to male gross tertiary enrollment
Female LFPR	SL.TLF.CACT.FE.ZS	Female labor force participation rate (%, 15+, ILO modeled)
Population	SP.POP.TOTL	Total population

All data from the World Bank Open Data platform, sourced from UNESCO Institute for Statistics (education) and ILO (labor force). CC BY 4.0.

Files:

File	Description
data/gpi_tertiary_enrollment.csv	Cross-section: 165+ economies, most recent GPI, region, income group, population
data/gpi_historical.csv	Panel: 197 countries, annual data (1970–2024), from World Bank API
data/gpi_vs_lfpr.csv	Cross-section: GPI + female LFPR for ~60 countries (all columns from World Bank API)

To refresh the data from the World Bank API:

python scripts/01_fetch_data.py    # GPI cross-section
python scripts/01b_fetch_historical.py  # Historical panel (all years)
python scripts/02_fetch_lfpr.py    # Female LFPR

No API key required. All data from World Bank Open Data.

Analysis

python scripts/03_analyze.py

Prints summary statistics, region and income decompositions, extremes, reversal analysis, and a big-country table. Exports data/gpi_viz_data.json.

Figures

python scripts/04_make_figures.py

Generates six figures in figs/:

Figure	Description
fig1_distribution.png	Histogram + population-scaled strip chart
fig2_regions.png	Population-weighted mean GPI by region
fig3_income.png	Population-weighted mean GPI by income group
fig4_big_countries.png	GPI dot plot for countries with pop > 50M
fig5_historical.png	Country trajectories 1970–2024 + share above parity over time
fig6_gpi_vs_lfpr.png	GPI vs. female labor force participation rate scatter

Repo structure

.
├── readme.md
├── requirements.txt
├── Makefile
├── data/
│   ├── gpi_tertiary_enrollment.csv
│   ├── gpi_historical.csv
│   └── gpi_vs_lfpr.csv
├── figs/
│   ├── fig1_distribution.png
│   ├── fig2_regions.png
│   ├── fig3_income.png
│   ├── fig4_big_countries.png
│   ├── fig5_historical.png
│   └── fig6_gpi_vs_lfpr.png
└── scripts/
    ├── 01_fetch_data.py       # Pull fresh GPI cross-section from World Bank API
    ├── 01b_fetch_historical.py # Pull full historical panel (all years) from World Bank API
    ├── 02_fetch_lfpr.py       # Update female LFPR in gpi_vs_lfpr.csv from World Bank API
    ├── 03_analyze.py          # Summary statistics and decompositions
    └── 04_make_figures.py     # Generate all 6 figures

Quick start

pip install -r requirements.txt
make all

Or run individual steps:

make data      # Fetch data from World Bank API
make analyze   # Run analysis
make figures   # Generate figures
make clean     # Remove generated files

Key findings

• 75% of countries have GPI > 1 (female > male tertiary enrollment)
• Historical reversal: In a balanced panel of 30 countries, the share with GPI > 1 rose from 33% (1990) to 67% (2024)
• Pop-weighted mean (1.1) is lower than the median (1.2)—the population giants lag
• Japan and South Korea are the only large high-income countries with GPI < 1
• MENA paradox: Several countries have GPI > 1 but overall female LFPR below 15%
• Income gradient is monotonic: low-income 0.7 → high-income 1.2

Requirements

pip install -r requirements.txt

Python 3.9+. 01_fetch_data.py and 02_analyze.py use only the standard library. 03_make_figures.py needs matplotlib and numpy.

License

Code: MIT. Data: World Bank Terms of Use (CC BY 4.0).