Reproductive Pressure

What Governments Ask of Women and What They Refuse to Give

Fast Facts for Busy Reviewers

Goal:
Understand how support systems, policy investments, and family structures interact with global fertility decline and whether government support is actually aligned with what families need.

Tools Used:
Tableau • Jupyter Notebooks • Excel • Python (Pandas, Statsmodels, Seaborn)

Key Wins:

👶 Fertility rates are still falling fast in countries like Japan and Korea, even with major investments
🍼 Countries with better-balanced support structures tend to have smaller fertility gaps
🌍 It’s not just how much governments spend it’s how the support is structured
📊 Most of the data was solid, but South Korea’s public spending had gaps. I flagged it early and handled it carefully so it wouldn’t throw things off.

Strategic Takeaways:

Break support into categories (cash, childcare, education, tax relief) instead of lumping everything together
Watch for steep fertility declines in places that are trying to raise birth rates the disconnect says a lot
Use data storytelling to shift the narrative from “why don’t people have kids?” to “why is the system built this way?”

Explore the Code Behind the Insights

GitHub

Tableau

The Challenge

Fertility rates are dropping around the world even in countries pouring billions into pro-natalist policies. That raised some big questions:

If governments are investing more, why aren’t people having more kids?
Why aren’t financial incentives working?
Is the system still built for a past that doesn’t exist anymore asking people to parent without support, rights, or safety?

I wanted to get under the surface of those numbers by breaking down what kinds of support exist (not just how much is spent), how they’re distributed, and what it all means for women and families in high-pressure policy environments.

Problem

Countries are throwing money at birth rates but most don’t know whether they’re targeting the real pain points for families. I wanted to track what’s actually being offered vs. what’s needed.

Tools Used

To explore trends and structure the analysis, I used:

Tableau
Excel
Python (Pandas, Seaborn, Statsmodels)
Jupyter Notebook

Key Insights

Countries like Japan and Korea are still seeing steep fertility drops, even with high spending pointing to deeper structural issues.
The U.S. spends less but offers more through direct tax relief and childcare structure matters more than totals.
Support systems built only around cash benefits miss the bigger picture: time, care, and autonomy matter too.

Business Suggestions

Break down support types (cash, childcare, tax, etc.) in policy design one-size-fits-all spending isn’t working
Rethink family-centered support as a long term investment, not a short term fix
Center women’s needs in future planning not just economic targets

Data Cleaning Highlights

Before diving into the analysis, I made sure the dataset was clean, aligned, and ready for action. Here's a quick look at what I cleaned up:

Standardized country names like “Korea” and “Republic of Korea” to match across datasets
Cleaned and reformatted columns into snake_case for clarity (e.g., public_spending_gdp)
Combined datasets on country and year to link spending, fertility, and policy info
Filled missing values for South Korea using OWID and verified with .isnull()
Ran Dickey-Fuller to prep time series for forecasting and model stability

🧬 Most of the data came through clean from the World Bank, OECD, and OWID, but South Korea’s public spending had gaps I couldn’t fully recover.

🧬 I used `.isnull()` to flag missing values early, cross-checked what I could, and left out anything that felt shaky. No point forecasting if half the story’s missing.

🧬 From mismatched country names to raw sequences without structure this before/after table shows how I got everything into shape for regression, clustering, and forecasting.

🧬 Including simple visuals like this to highlight transformations in a clear way no need to guess what changed behind the scenes.

Snapshot of Key Cleaning Improvements

Want to see the original data?

➔ World Bank • OECD • OWID

Analytical Build

After cleaning and reshaping the data, it still wasn’t enough to just compare numbers. To really understand what was going on, support had to be restructured, new features had to be created, and deeper methods had to be used to connect the dots.

🧬 Engineered Features

Support traits like cash benefits, childcare, education, services, tax breaks, and fertility divergence were combined into a single score to make comparisons possible. Fertility Divergence showed how far each country’s fertility rate stood from the global average. Policy traits were also structured to set up clustering later on.

🧬 Clustering Analysis

Countries were grouped using KMeans based on how they structured support, not just how much they spent. The clusters showed that more spending did not always mean higher fertility. Structure mattered more than totals.

🧬 Time Series Analysis

Stationarity checks, ACF plots, and decomposition helped break down fertility trends over time. Japan’s trend showed a steady decline even as support increased, suggesting cultural and economic pressures that policy alone hasn’t solved.

These engineered features made the deeper analysis possible. Without them, the insights wouldn’t have had the same clarity or impact.

Want to see the full build in action? Check out the full Jupyter Notebooks here:
➔ Reproductive Pressure: Notebooks on GitHub

Key Visuals & What They Show

🧬 Fertility Pressure Map

Fertility is lowest where women face the most pressure and the least support. This map uses a custom Fertility Pressure Score that combines policy support levels and average fertility rate. Countries like South Korea (0.72), Spain (1.21), and Italy (1.28) remain far below average despite major policy efforts. Mexico (1.91) and New Zealand (1.67) show stronger rates with less spending, suggesting it's not just about money it's also about structure.

🧬 Fertility Divergence Bar

Fertility Divergence measures how far a country’s fertility rate is from the global average (around 2.0). A negative value means it’s falling behind; a positive one means it’s holding steady or improving. Countries like Mexico (+0.4) and New Zealand (+0.2) are moving above average, while places like Italy (–0.3) and Lithuania (–0.3) are drifting further behind. Support alone isn’t enough structure, access, and cultural context all shape outcomes.

🧬 Fertility Over Time

This line chart shows fertility trends for Japan, Korea, and the U.S. over time. Japan and Korea have dropped well below 1.0, while the U.S. is holding just above 1.6. These aren't just policy outcomes they reflect decades of cultural, economic, and structural change. The data makes one thing clear: support has grown, but it hasn’t been enough to shift the trend.

🧬 What Support Actually Looks Like

Support-rich countries topped the charts in cash (13.2), childcare (10.2), and family services (15.0), but still had low fertility. This grouped bar chart breaks support into six traits to show how each policy environment is structured not just how much is spent. The low-fertility/modest-support group hovered around 3.5 in childcare and 4.1 in services. The takeaway: access matters just as much as investment.

Want to see even more?
Take a closer look at the full Tableau dashboard that ties these visuals together

➔ Explore the Full Story Here

Main Insights (Recap)

📉 Time Doesn’t Heal the Gap

Even with rising investment, Japan and Korea’s fertility rates kept dropping.
Japan’s spending increased from 1.2% to 1.7% of GDP, and Korea went from 0.7% to 3.4%, but both countries still fell to around 1.0 births per woman. The U.S. spent less and declined more gradually but it declined all the same. This isn’t just about how much is spent. It’s about timing, design, and whether support actually reaches people in time to matter.

⚖️ Equality on Paper ≠ Equality in Practice

The Gender Inequality Index made one thing clear: support isn’t enough if women are still locked out of opportunity. In countries with generous benefits, women still faced wide gaps in workforce participation, pay, and leadership. Fertility doesn’t rise when people feel like they’re risking their future to have a child even if the benefits are there on paper.

🛡️ Rights Without Support ≠ Real Choice

Having legal access to abortion doesn’t guarantee real support and having support doesn’t guarantee autonomy. Some countries recognize reproductive rights but still offer little in the way of care, time, or protection. Others spend more but restrict choice. Without both freedom and support, parenthood starts to feel like a mandate not a choice.

💸 What It Costs to Care

Out-of-pocket costs are still shaping the choice to parent. In two-parent households, net childcare costs can eat up 30% of income. In some countries, single parents pay over 40% just to access care. Even with strong systems in place, the cost of care is still pushing families away from having more children or any at all.

🧮 Structure Behind the Numbers

Regression didn’t explain much, so clustering and time series analysis stepped in. Countries were grouped by how support was structured, not how much was spent and the gaps became clear. Japan’s fertility kept falling even with more support. The insight? Structure, timing, and design matter more than totals.

From Data to Decisions

This project uncovered how government policies, support structures, and social expectations shape reproductive choices. The analysis didn’t just look at fertility rates it dug into why they’re falling, where support breaks down, and how pressure systems push people to delay or avoid parenthood altogether.

Hands-on tools and techniques included:

Cleaning and merging multi-source datasets (World Bank, OECD, OWID) to build a full view across countries
Using unsupervised clustering and supervised regression to group countries by support structure and test policy impact
Applying geospatial and time series analysis to map global fertility pressures and break down trends over time
Visualizing fertility, spending, and support patterns to spot where structure matters more than totals

Business Recommendations:

01: Redesign Family Policy Around Structure, Not Size

Countries like Japan, Korea, and the U.S. increased spending after 2010, but fertility still dropped. Japan’s spending rose from 1.2% to 1.7% of GDP, and Korea went from 0.7% to 3.4%, but both still fell to around 1.0 births per woman. Cash alone didn’t shift the trend. Countries with better-balanced investments across childcare, education, and services performed better over time.

➔ Policy should prioritize delivery and design, not just budget totals.

02: Make childcare affordability a core metric

Even in high-spending countries like Sweden (1.56) and France (1.26), childcare costs remained a major burden. It’s not about how much the government spends it’s about how much families still have to pay. In some places, net childcare costs can eat up to 30% of income for two-parent households and over 40% for single parents.

➔ Future policy should focus on reducing out-of-pocket costs, especially for single parents and low- to mid-income households.

03: Use clustering to expose hidden policy gaps

National averages can mask serious disconnects.
Clustering analysis showed that countries with similar support totals didn’t mean they were getting better results.

Cluster 0 (moderate support, Total Benefits around 2.85),
Cluster 1 (lowest support, Total Benefits around 1.44), and
Cluster 2 (strongest support, Total Benefits around 3.23).

all had different fertility outcomes, even when spending levels looked close.

➔ Use cluster segmentation to find underperformers especially places with strong policy on paper but poor results in practice.

04: Link benefits to rights and safety

Countries with stronger reproductive rights and legal protections saw more stable fertility patterns. Without legal autonomy and basic safety, no amount of support feels usable.

➔ Support only matters if people feel safe choosing when and how to start a family.

05: Design for long-term stability, not one-time incentives

Short-term bonuses can’t fix structural issues. Countries like Mexico and New Zealand offered lean but well-balanced systems and it worked.

➔ Build multi-layered, sustainable support systems that grow with families, not against them.