deveconomics-toolkit

Interactive Shiny apps for development economics, impact evaluation, and the international development sector.

A suite of 11 open-source tools built with R Shiny and Python Shiny — designed for researchers, practitioners, and students working in global development. Created for Impact Mojo.

---

The Apps

Impact Evaluation & Econometrics

App	Framework	What it does
RCT Power Calculator	Python Shiny	Compute sample size, statistical power, and MDE for randomized controlled trials. Supports cluster randomization and multiple treatment arms.
DiD Simulator	Python Shiny	Simulate difference-in-differences designs with OLS regression output, event-study plots, and parallel trends violation detection.
RDD Explorer	R Shiny	Explore sharp and fuzzy regression discontinuity designs with local polynomial estimation, McCrary manipulation tests, and bandwidth sensitivity.
Synthetic Control Visualizer	Python Shiny	Visualize the synthetic control method with constrained optimization, in-space placebo tests, and pre-treatment fit diagnostics.

Poverty & Inequality

App	Framework	What it does
Gini & Lorenz Curve Tool	R Shiny	Calculate Gini, Theil, and Palma ratio from generated or user-supplied data. Compare distributions side by side with interactive Lorenz curves.
MPI Explorer	Python Shiny	Explore the Multidimensional Poverty Index using the Alkire-Foster method. Radar charts, decomposition by dimension, and sensitivity to the poverty cutoff.
Poverty Line Analysis	R Shiny	Compute FGT(0/1/2) and Watts indices against World Bank poverty lines ($2.15, $3.65, $6.85). Simulate growth and redistribution scenarios.

Program Design & Management

App	Framework	What it does
Theory of Change Visualizer	Python Shiny	Build interactive ToC diagrams with sector templates (Education, Health, Livelihoods, WASH). Auto-generates narrative descriptions and exportable summaries.
Cost-Benefit Analysis Tool	Python Shiny	Calculate NPV, BCR, and IRR with tornado diagrams, Monte Carlo simulation, and a cost-effectiveness tab benchmarked against GiveWell thresholds.
LogFrame Builder	R Shiny	Build logical frameworks with editable matrices, indicator tracking with progress bars, results chain visualization, and HTML/CSV export. 6 sector templates.

Data Exploration

App	Framework	What it does
WDI Dashboard	R Shiny	Explore 15 development indicators across 50 countries and 6 regions. Time series, cross-country comparisons, scatter plots, and downloadable data tables.

---

Quick Start

Python Shiny apps

# Pick any Python app
cd python-shiny/rct-power-calculator

# Create a virtual environment (recommended)
python -m venv .venv
source .venv/bin/activate   # Linux/Mac
# .venv\Scripts\activate    # Windows

# Install dependencies and run
pip install -r requirements.txt
shiny run app.py

The app will be available at http://127.0.0.1:8000.

R Shiny apps

# From R or RStudio — pick any R app
shiny::runApp("r-shiny/rdd-explorer")

Or from the command line:

cd r-shiny/rdd-explorer
Rscript -e "shiny::runApp('.')"

The app will be available at http://127.0.0.1:3838 (or the port R assigns).

R dependencies

All R apps use standard CRAN packages. Install them once:

install.packages(c("shiny", "ggplot2", "dplyr", "tidyr", "DT"))

---

Project Structure

deveconomics-toolkit/
├── README.md
├── LICENSE
├── .gitignore
│
├── python-shiny/
│   ├── rct-power-calculator/
│   │   ├── app.py
│   │   └── requirements.txt
│   ├── did-simulator/
│   │   ├── app.py
│   │   └── requirements.txt
│   ├── mpi-explorer/
│   │   ├── app.py
│   │   └── requirements.txt
│   ├── theory-of-change/
│   │   ├── app.py
│   │   └── requirements.txt
│   ├── cost-benefit-analysis/
│   │   ├── app.py
│   │   └── requirements.txt
│   └── synthetic-control/
│       ├── app.py
│       └── requirements.txt
│
└── r-shiny/
    ├── rdd-explorer/
    │   └── app.R
    ├── gini-lorenz/
    │   └── app.R
    ├── poverty-line-analysis/
    │   └── app.R
    ├── logframe-builder/
    │   └── app.R
    └── wdi-dashboard/
        └── app.R

---

Testing

pip install -r requirements-dev.txt
python -m pytest tests/ -q

CI compiles every Python app, imports each one and asserts it defines a Shiny App, runs the test suite, and parses every R app.

On the power calculator's numbers

All three solve modes were wrong until 2026-09-22, each in the direction that flatters a study. The cause was a missing factor of two from Var(mean_treat - mean_control) = sigma^2/n_t + sigma^2/n_c.

solve mode	was reported	correct
n per arm at d = 0.2, 80% power	197	393
power at 393 per arm	97.8%	80.0%
MDE at 393 per arm	0.141 SD	0.200 SD

Simulated, the old recommendation of 197 per arm delivers 51% power. n_total was correct throughout, which is why the error survived: the one figure anyone would check against a published table was the one that was right.

The arithmetic now lives in python-shiny/rct-power-calculator/power.py and is checked against both closed form and a simulated t-test in tests/test_power.py.

Known limitation, stated rather than buried. The module uses the normal approximation rather than the t distribution, so it understates the required sample slightly at small n: 63 per arm at d = 0.5 where GPower asks for 64, and 25 at d = 0.8 where GPower asks for 26. Simulated, those land at 79.6% and 79.1% power against a nominal 80%. The gap is under one percentage point above roughly 25 per arm and widens below it. A test pins its size.

Checked and found correct, on the same date and by the same standard: the Alkire-Foster implementation in mpi-explorer, and the NPV, benefit-cost ratio, IRR and discounted-payback functions in cost-benefit-analysis, which agree with closed form exactly.

The five R apps have a parse check and no test coverage.

Deployment Options

Each app is self-contained and can be deployed independently:

Platform	Best for	Guide
shinyapps.io	R apps — free tier available	rsconnect::deployApp("r-shiny/rdd-explorer")
Hugging Face Spaces	Python apps — free GPU/CPU	Create a Space with the Gradio/Docker SDK and point to the app
Shinylive	Python apps — runs entirely in the browser via WebAssembly	shinylive export app.py site/
Posit Connect	Enterprise deployment for both R and Python	Follow Posit Connect docs
Docker	Any app — portable containers	See Dockerfiles (coming soon)
Your website	Embed any deployed app via <iframe>	<iframe src="https://your-app-url" width="100%" height="800px"></iframe>

---

Tech Stack

• Python Shiny (6 apps) — shiny, numpy, scipy, matplotlib, pandas, statsmodels
• R Shiny (5 apps) — shiny, ggplot2, dplyr, tidyr, DT
• All apps generate synthetic data for demonstration — no external API calls or datasets required

---

Key References

These apps implement methods from the development economics canon:

• Angrist, J. & Pischke, J. (2009). Mostly Harmless Econometrics
• Duflo, E., Glennerster, R. & Kremer, M. (2007). Using Randomization in Development Economics Research
• Abadie, A., Diamond, A. & Hainmueller, J. (2010). Synthetic Control Methods
• Imbens, G. & Lemieux, T. (2008). Regression Discontinuity Designs
• Alkire, S. & Foster, J. (2011). Counting and Multidimensional Poverty Measurement
• Foster, J., Greer, J. & Thorbecke, E. (1984). A Class of Decomposable Poverty Measures
• Card, D. & Krueger, A. (1994). Minimum Wages and Employment

---

Contributing

Contributions are welcome! If you'd like to add a new app or improve an existing one:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/new-app)
3. Add your app in the appropriate directory (python-shiny/ or r-shiny/)
4. Include a requirements.txt (Python) or document R dependencies
5. Submit a pull request

---

License

MIT License — see LICENSE for details.

---

Built with care for the global development community.