Government System Type and Education Enrollment in Africa

A World Bank Education Enrollment Analysis

Introduction

Education access is one of the most important indicators of long-term development. Countries with higher enrollment rates are generally better positioned to improve human capital, expand economic opportunity, and reduce inequality. However, education systems are shaped not only by income and geography, but also by political institutions.

This report examines whether education enrollment differs across African countries with different types of government systems. Specifically, the analysis compares countries categorized as presidential, parliamentary, and semi-presidential systems. The main outcome variables are gross enrollment rates at the primary, secondary, and tertiary education levels.

The central research question is:

How does education enrollment vary by government system type across African countries?

This question is analyzed using cleaned education enrollment data for African countries. The report focuses on both cross-sectional differences between government system types and time trends in enrollment from 2010 onward.

Data Description

The dataset used in this project contains education enrollment data for African countries. The main cleaned dataset is education_africa_cleaned.csv. It includes country-level information such as country name, government type, country code, region, income group, year, and gross enrollment rates for primary, secondary, and tertiary education.

The key variables are:

Variable Meaning
country Name of the African country
government Detailed government system classification
country_code Three-letter country code
region Regional classification
income_group World Bank income group
year Calendar year
school_enrollment,_primary_%_gross Gross primary enrollment rate
school_enrollment,_secondary_%_gross Gross secondary enrollment rate
school_enrollment,_tertiary_%_gross Gross tertiary enrollment rate

Gross enrollment rates can exceed 100% because they include students who are younger or older than the official age group for that education level.

Setup

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import sqlite3
from pathlib import Path

sns.set_theme(style="whitegrid")
plt.rcParams["figure.figsize"] = (12, 7)

Load Data

# Load cleaned education dataset
data_path = Path("data/clean-data/education_africa_cleaned.csv")

df = pd.read_csv(data_path)

df.head()
country government country_code region income_group year school_enrollment,_primary_%_gross school_enrollment,_secondary_%_gross school_enrollment,_tertiary_%_gross
0 Algeria semi-presidential unitary republic DZA Middle East & North Africa Upper middle income 1971.0 74.083488 12.10871 NaN
1 Algeria semi-presidential unitary republic DZA Middle East & North Africa Upper middle income 1972.0 79.225311 12.65980 NaN
2 Algeria semi-presidential unitary republic DZA Middle East & North Africa Upper middle income 1973.0 85.245468 14.07465 NaN
3 Algeria semi-presidential unitary republic DZA Middle East & North Africa Upper middle income 1974.0 90.666229 15.19896 NaN
4 Algeria semi-presidential unitary republic DZA Middle East & North Africa Upper middle income 1975.0 94.406937 16.36799 NaN 
df.shape
(2303, 9)
df.info()
<class 'pandas.DataFrame'>
RangeIndex: 2303 entries, 0 to 2302
Data columns (total 9 columns):
 #   Column                                Non-Null Count  Dtype  
---  ------                                --------------  -----  
 0   country                               2303 non-null   str    
 1   government                            2303 non-null   str    
 2   country_code                          2303 non-null   str    
 3   region                                2303 non-null   str    
 4   income_group                          2303 non-null   str    
 5   year                                  2303 non-null   float64
 6   school_enrollment,_primary_%_gross    2193 non-null   float64
 7   school_enrollment,_secondary_%_gross  1689 non-null   float64
 8   school_enrollment,_tertiary_%_gross   1513 non-null   float64
dtypes: float64(4), str(5)
memory usage: 162.1 KB

Data Cleaning

The original government variable contains detailed descriptions, such as “presidential republic,” “semi-presidential republic,” and “parliamentary republic.” For this analysis, these detailed categories are simplified into three broader government system types:

  1. Presidential
  2. Parliamentary
  3. Semi-presidential

This makes the results easier to interpret and aligns the analysis with the figures comparing education outcomes by system type.

# Standardize column names for easier coding
df = df.rename(columns={
    "school_enrollment,_primary_%_gross": "primary_enrollment",
    "school_enrollment,_secondary_%_gross": "secondary_enrollment",
    "school_enrollment,_tertiary_%_gross": "tertiary_enrollment"
})

# Create simplified government system type
def classify_system(gov):
    gov = str(gov).lower()
    if "semi-presidential" in gov:
        return "semi-presidential"
    elif "presidential" in gov:
        return "presidential"
    elif "parliamentary" in gov:
        return "parliamentary"
    else:
        return "other"

df["system_type"] = df["government"].apply(classify_system)

# Keep only the three main government system types
df = df[df["system_type"].isin(["presidential", "parliamentary", "semi-presidential"])].copy()

df[["country", "government", "system_type", "year", "primary_enrollment", "secondary_enrollment", "tertiary_enrollment"]].head()
country government system_type year primary_enrollment secondary_enrollment tertiary_enrollment
0 Algeria semi-presidential unitary republic semi-presidential 1971.0 74.083488 12.10871 NaN
1 Algeria semi-presidential unitary republic semi-presidential 1972.0 79.225311 12.65980 NaN
2 Algeria semi-presidential unitary republic semi-presidential 1973.0 85.245468 14.07465 NaN
3 Algeria semi-presidential unitary republic semi-presidential 1974.0 90.666229 15.19896 NaN
4 Algeria semi-presidential unitary republic semi-presidential 1975.0 94.406937 16.36799 NaN 
# Check number of observations by system type
df["system_type"].value_counts()
system_type
presidential         1570
semi-presidential     498
parliamentary         235
Name: count, dtype: int64
# Check number of countries by system type
df.groupby("system_type")["country"].nunique().reset_index(name="number_of_countries")
system_type number_of_countries
0 parliamentary 5
1 presidential 36
2 semi-presidential 11

SQL Descriptive Statistics

The project uses SQL for descriptive statistics. The cleaned dataset is loaded into a SQLite database, and SQL queries are used to summarize average education enrollment by government system type.

conn = sqlite3.connect(":memory:")
df.to_sql("education", conn, index=False, if_exists="replace")
2303
query = """
SELECT
    system_type,
    COUNT(*) AS observations,
    COUNT(DISTINCT country) AS countries,
    ROUND(AVG(primary_enrollment), 2) AS avg_primary_enrollment,
    ROUND(AVG(secondary_enrollment), 2) AS avg_secondary_enrollment,
    ROUND(AVG(tertiary_enrollment), 2) AS avg_tertiary_enrollment
FROM education
GROUP BY system_type
ORDER BY avg_secondary_enrollment DESC;
"""

summary_by_system = pd.read_sql_query(query, conn)
summary_by_system
system_type observations countries avg_primary_enrollment avg_secondary_enrollment avg_tertiary_enrollment
0 parliamentary 235 5 99.84 45.10 11.03
1 semi-presidential 498 11 85.44 41.94 11.78
2 presidential 1570 36 84.57 29.60 4.90 
query = """
SELECT
    system_type,
    MIN(year) AS first_year,
    MAX(year) AS last_year,
    ROUND(AVG(primary_enrollment), 2) AS mean_primary,
    ROUND(AVG(secondary_enrollment), 2) AS mean_secondary,
    ROUND(AVG(tertiary_enrollment), 2) AS mean_tertiary
FROM education
GROUP BY system_type;
"""

pd.read_sql_query(query, conn)
system_type first_year last_year mean_primary mean_secondary mean_tertiary
0 parliamentary 1970.0 2024.0 99.84 45.10 11.03
1 presidential 1970.0 2025.0 84.57 29.60 4.90
2 semi-presidential 1970.0 2025.0 85.44 41.94 11.78

Exploratory Analysis

Enrollment Distribution by Government System Type

To compare education enrollment across government types, the first figure uses boxplots for primary, secondary, and tertiary enrollment.

# Convert to long format for plotting
plot_df = df.melt(
    id_vars=["country", "year", "system_type"],
    value_vars=["primary_enrollment", "secondary_enrollment", "tertiary_enrollment"],
    var_name="education_level",
    value_name="enrollment"
)

plot_df["education_level"] = plot_df["education_level"].replace({
    "primary_enrollment": "Primary",
    "secondary_enrollment": "Secondary",
    "tertiary_enrollment": "Tertiary"
})

# Drop missing values
plot_df = plot_df.dropna(subset=["enrollment"])

plt.figure(figsize=(12, 7))
sns.boxplot(
    data=plot_df,
    x="education_level",
    y="enrollment",
    hue="system_type",
    order=["Primary", "Secondary", "Tertiary"],
    hue_order=["semi-presidential", "presidential", "parliamentary"]
)

plt.title("Education Enrollment by Government Type in Africa")
plt.xlabel("Education Level")
plt.ylabel("Enrollment (%)")
plt.legend(title="system_type")
plt.tight_layout()
plt.show()

The boxplot shows that primary education enrollment is high across all three government system types. This suggests that primary education access has become relatively widespread across the countries included in the dataset. However, secondary and tertiary enrollment show much larger differences.

Parliamentary systems appear to have higher median secondary and tertiary enrollment than presidential systems. Semi-presidential systems also perform relatively well in some years, but they appear more variable. Presidential systems tend to have lower secondary and tertiary enrollment, especially at the tertiary level.

Additional Summary Tables

# Average enrollment by government system type since 2010
recent_summary = recent_df.groupby("system_type").agg(
    countries=("country", "nunique"),
    observations=("country", "count"),
    avg_primary=("primary_enrollment", "mean"),
    avg_secondary=("secondary_enrollment", "mean"),
    avg_tertiary=("tertiary_enrollment", "mean")
).round(2).reset_index()

recent_summary
system_type countries observations avg_primary avg_secondary avg_tertiary
0 parliamentary 4 57 105.04 79.22 23.99
1 presidential 36 404 102.18 49.60 11.33
2 semi-presidential 11 124 101.47 62.82 23.04 
# Most recent available average by system type
latest_year = int(df["year"].max())

latest_df = df[df["year"] == latest_year]

latest_summary = latest_df.groupby("system_type").agg(
    countries=("country", "nunique"),
    avg_primary=("primary_enrollment", "mean"),
    avg_secondary=("secondary_enrollment", "mean"),
    avg_tertiary=("tertiary_enrollment", "mean")
).round(2).reset_index()

latest_year, latest_summary
(2025,
          system_type  countries  avg_primary  avg_secondary  avg_tertiary
 0       presidential          2       100.53          23.13          6.38
 1  semi-presidential          2       105.48            NaN         38.04)

Results

The analysis shows three main findings.

First, primary education enrollment is high across all government system types. This suggests that primary schooling has expanded broadly across African countries, regardless of whether a country has a presidential, parliamentary, or semi-presidential system.

Second, larger differences appear at the secondary education level. Parliamentary systems generally have higher secondary enrollment than presidential systems. Semi-presidential systems often fall between parliamentary and presidential systems, though they show greater year-to-year variation.

Third, tertiary enrollment is the lowest of the three education levels, but it shows the clearest long-term difference between government types. Parliamentary and semi-presidential systems generally have higher tertiary enrollment than presidential systems. Presidential systems remain lower throughout most of the observed period.

Overall, the results suggest that government system type is associated with differences in education enrollment, especially beyond the primary level. However, this relationship should not be interpreted as causal. Government type may be correlated with other important factors, including income level, colonial history, political stability, public spending, and state capacity.

Limitations

This project has several limitations.

First, the analysis is descriptive and does not prove that government system type causes higher or lower education enrollment. The observed differences may be influenced by income, region, population size, conflict, or historical factors.

Second, the dataset contains missing values, especially for secondary and tertiary education. Because tertiary education has the least complete data, trends at that level should be interpreted cautiously.

Third, the government system categories are simplified. Many countries have more specific institutional arrangements, such as federal presidential republics, unitary presidential republics, constitutional monarchies, or hybrid systems. Grouping them into only three categories makes the analysis easier to interpret, but it may hide important variation.

Fourth, the analysis uses gross enrollment rates. Gross enrollment can exceed 100%, so it should not be interpreted as the exact percentage of children of the official age group enrolled in school.

Conclusion

This report examined education enrollment in African countries by government system type. The analysis found that primary enrollment is high across all systems, but secondary and tertiary enrollment differ more strongly by government type.

Parliamentary systems generally show higher secondary and tertiary enrollment than presidential systems. Semi-presidential systems also show relatively strong outcomes in some years, especially for tertiary education, but with more volatility.

The findings suggest that political institutions may be connected to educational outcomes, particularly at higher levels of education. However, future research should include additional controls such as income group, public education spending, conflict status, and regional effects before making stronger causal claims.

Group Contributions

The project repository was managed using GitHub. Contributions were evaluated using GitHub commit history, additions, deletions, and each group member’s role in the project.

git shortlog -sn --all
git log --all --format='%aN' | sort -u | while read name; do
  echo "Author: $name"
  git log --all --author="$name" --pretty=tformat: --numstat | \
    awk '{ added += $1; deleted += $2 } END { print "Lines added:", added, "Lines deleted:", deleted, "Net lines:", added - deleted }'
  echo
done

References

World Bank. (2024). World Development Indicators. World Bank.

Project dataset: education_africa_cleaned.csv.

Project codebook: CODEBOOK.md.

Project cleaning script: clean_joined_data.py.