$180.9B
Total Committed
1,319
Loan Deals
49
African Nations
0
Mega-Deals Since 2019

I. The Boom and Bust

Chinese lending followed a dramatic arc. But the headline numbers hide a deeper story. The 2016 peak of $28.8B was driven by just two mega-deals worth $14.1B — nearly half the year’s total. Strip those out and 2016 looks ordinary.

Annual Lending Volume ($B)
The red bar marks 2016 — 49% of that year came from just 2 deals over $1B each.
Mega-Deals (>$1B) vs Everything Else
Billion-dollar deals vanished entirely after 2018. Not a single mega-deal in 6 years.

The “peak” was never about more lending — it was about bigger individual bets. When those bets stopped, the system contracted.

↓

II. The Disappearing Small Loan — Then a Surprise Return

In 2000, 92% of loans were under $20M. By 2020, just 5%. Then in 2024, small loans surged back to 45% — a fundamental shift in lending philosophy.

Share of Small Loans (<$20M) per Year
The U-shaped curve: China is pivoting from mega-infrastructure to smaller, targeted investments.
92%
Small Loans 2000
5%
Small Loans 2020
45%
Small Loans 2024

China is not “leaving Africa” — it’s changing how it lends. The next section shows where the money concentrated.

↓

III. One Country, One Sector: Hidden Debt Traps

The truly alarming pattern: most top borrowers funnelled the majority of their Chinese debt into just one industry. If that sector falters, the entire debt structure is at risk.

Top 15 Borrowers ($B)
Angola’s $49.2B is 3.5× the next borrower.
Single-Sector Dependency (%)
South Africa: 100% coal. Uganda: 82% hydro. Angola: 65% oil.

“South Africa borrowed $4.5B from China — every dollar went to coal. As the world decarbonises, this is a stranded-asset time bomb.”

Pattern: sector-country cross-tabulation
Debt Bunching: Top 3 Years as % of Total Borrowing
56–67% of each country’s debt was signed in just 3 years — enormous refinancing risk.
↓

IV. The Quiet Power Shift Between Two Banks

In 2016, Exim Bank’s share collapsed to just 19%. CDB and syndicated lenders drove the surge. When they pulled back, volumes cratered.

Exim Bank’s Share of Volume (%)
97% in 2007 → 19% in 2016 → 83% in 2024, but of a much smaller pie.
Avg Deal Size: Exim vs CDB ($M)
CDB averaged 2–3× larger. In 2018: CDB $966M vs Exim $159M.

“CDB is the quiet giant. When it stopped writing cheques, African lending volumes cratered.”

Pattern: lender-level time series decomposition
↓

V. After COVID: A Completely Different Portfolio

Oil and coal financing dropped to zero. Solar is the only energy source that grew. No new borrowing countries since 2015.

Energy Shift: Pre vs Post 2020 (% of Volume)
Oil: 13% → 0%. Coal: 7% → 0%. Solar doubled.
New Borrowing Countries per Year
Zero new countries since 2015. The market was saturated before BRI matured.
$0
Oil Post-2020
$0
Coal Post-2020
+109%
Solar Growth
0
New Countries Since 2015

China hasn’t abandoned Africa — it has reinvented its lending.

↓

VI. What the Algorithms Revealed

A country’s borrowing history is the strongest predictor of new loan size — not sector, year, or income level. Past lending begets future lending.

What Predicts Loan Size? (Feature Importance)
Country Total Volume (28%) dominates. Year and sector barely register.
Borrower Clusters (K-Means)
4 types: mega-borrowers, mid-tier, occasional, one-off.
Classification Accuracy by Model
Gradient Boosting: 91% accuracy on loan categories.
Lending Inequality (Gini Coefficient)
Inequality surged from 0.4 to 0.8+ during peak years.
↓

VII. What Comes Next?

Polynomial trend confirms the peak has passed. Lending will stabilise at $2–5B/year — 85–90% below peak.

Historical Volume & Trend
The parabolic curve peaked around 2015–16 and shows no recovery.

The Six Hidden Patterns

What Traditional Analysis Missed

  1. The mega-deal mirage. 2016’s “peak” was 2 deals worth $14.1B. No billion-dollar deal since 2018.
  2. The small-loan comeback. Under-$20M loans: 92% (2000) → 5% (2020) → 45% (2024).
  3. Single-sector debt traps. South Africa: 100% coal. Uganda: 82% hydro. Angola: 65% oil.
  4. CDB was the real engine. Deals 2–3× larger than Exim. Its withdrawal cratered volumes.
  5. No new countries since 2015. Market saturated five years before COVID.
  6. History predicts the future. Borrowing history is 5× more predictive than sector or year.