Net Financial Flows to Developing Countries

Data and code for The Great Reversal report.

Development Finance Observatory | ONE DATA

---

Overview

This project analyzes net financial flows to low and lower-middle income countries by combining:

• Official Development Assistance (ODA) and Other Official Flows data from OECD DAC
• Total Official Support for Sustainable Development (TOSSD)
• Debt statistics from World Bank International Debt Statistics

Data Sources

Source	Provider	What It Contains
OECD DAC/CRS	OECD	ODA grants, loans, equity investments, other official flows
TOSSD	International Forum on TOSSD	TOSSD
International Debt Statistics	World Bank	PPG long-term external debt disbursements and debt service

Key Concepts

Net flows = Gross inflows − Debt service payments

• PPG debt: Public and Publicly Guaranteed long-term external debt
• Country coverage: Low Income Countries (LIC) and Lower-Middle Income Countries (LMIC) only
• Excluded: Upper-middle income countries, high income countries
• Outliers excluded: China, Libya, Ukraine, Venezuela (as recipients)
• Special cases: Ethiopia classified as LIC throughout
• Prices: Constant 2024 USD using pydeflate

Quick Start

# Prerequisites: Python 3.13+, uv package manager
git clone https://github.com/ONEcampaign/DFO-Net-Flows.git
cd DFO-Net-Flows
uv sync

# Generate the analysis
uv run python scripts/analysis/net.py
uv run python scripts/analysis/replication.py

# Output: output/replication_analysis.xlsx

Data Pipeline

┌─────────────────────────────────────────────────────────────┐
│                     DATA SOURCES                            │
├─────────────────────────────────────────────────────────────┤
│  OECD DAC/CRS      TOSSD            World Bank IDS          │
│  (oda.py)          (tossd.py)       (debt.py)               │
└──────────┬─────────────┬────────────────┬───────────────────┘
           │             │                │
           ▼             ▼                ▼
┌─────────────────────────────────────────────────────────────┐
│                    ANALYSIS LAYER                           │
├─────────────────────────────────────────────────────────────┤
│  inflows.py → Combines grants + debt + other disbursements  │
│  outflows.py → Debt service payments (principal + interest) │
│  net.py → Merges inflows/outflows + adds metadata           │
└──────────────────────────┬──────────────────────────────────┘
                           │
                           ▼
┌─────────────────────────────────────────────────────────────┐
│                    OUTPUT FILES                             │
├─────────────────────────────────────────────────────────────┤
│  net_flows_agg.parquet     - Main dataset                   │
│  replication_analysis.xlsx - Report figures with data       │
└─────────────────────────────────────────────────────────────┘

Replicating Report Figures

The replication.py module provides statement-driven analysis that matches claims from The Great Reversal report. Each statement generates an Excel sheet with:

• Title and narrative explaining the finding
• Backing data that supports the claim

Run uv run python scripts/analysis/replication.py to generate all statement sheets.

Output Files

File	Description
inflows_by_source_and_type.parquet	Inflows by source (DAC, TOSSD, IDS)
outflows_by_source_and_type.parquet	Debt service payments
net_flows_agg.parquet	Combined flows with country metadata
replication_analysis.xlsx	Report figures with backing data

For Researchers: Custom Analysis

Use the research module for flexible queries:

from scripts.research.data import get_flows
from scripts.research.aggregate import by_year, by_period, totals

# Get all LIC/LMIC flows (excludes outliers by default)
data = get_flows()

# Filter by region and time
africa = get_flows(regions=["Africa"], years=(2020, 2024))

# Filter by counterpart type
bilateral = get_flows(counterpart_types=["Bilateral"])

# Filter by debt distress risk
high_risk = get_flows(debt_distress=["High", "In Distress"])

# Aggregate by year
yearly = by_year(data)

# Compare periods
periods = by_period(data, periods=[(2010, 2014), (2020, 2024)])

Available filters:

Parameter	Description	Example
years	Time range tuple	(2020, 2024)
income_levels	Income classification	["LIC", "LMC"]
regions	Geographic region	["Africa", "Asia"]
counterpart_types	Source type	["Bilateral", "Multilateral", "Private"]
countries	Specific countries	["Kenya", "Nigeria"]
debt_distress	DSA risk level	["High", "In Distress"]
include_outliers	Include China, Libya, etc.	True

Project Structure

DFO-Net-Flows/
├── scripts/
│   ├── sources/          # Data fetching
│   │   ├── oda.py        # OECD DAC/CRS data
│   │   ├── debt.py       # World Bank IDS
│   │   ├── tossd.py      # TOSSD data
│   │   └── metadata.py   # Income levels, DSA risk
│   ├── analysis/         # Processing pipeline
│   │   ├── inflows.py    # Combine inflow sources
│   │   ├── outflows.py   # Debt service payments
│   │   ├── net.py        # Main pipeline
│   │   └── replication.py # Report figures
│   ├── research/         # Researcher-facing API
│   │   ├── data.py       # get_flows() function
│   │   ├── aggregate.py  # by_year, by_period, totals
│   │   └── metrics.py    # Shares, ratios, growth
│   └── config.py         # Configuration
├── raw_data/             # Input data files
├── output/               # Generated outputs
└── pyproject.toml        # Dependencies

Dependencies

Key libraries:

Library	Purpose
oda-data	OECD DAC/CRS data access
bblocks-data-importers	World Bank IDS data
pydeflate	GDP deflation for constant prices
pandas	Data manipulation

Install with: uv sync

Data Limitations

• TOSSD: 2024 data interpolated from 2020-2023 average (publication lag)
• Chinese development finance: Based on debt service data; AidData commitment tracking captures different flows
• Not included: Philanthropic flows, remittances, illicit financial flows
• Not covered: Private capital flows (FDI, portfolio investment)

Development

# Lint and format
uv run ruff check scripts/
uv run ruff format scripts/

# Type checking
uv run mypy scripts/

# Run all pre-commit hooks
pre-commit run --all-files

Citation

Development Finance Observatory | ONE Data (2025). The Great Reversal.

Contact

Questions or feedback? Open an issue on GitHub or contact ONE Data.