Prerequisites

• Python 3.11+ with pip
• Git
• Free API keys: FRED and BEA

1. Clone the repository

git clone https://github.com/jonsmirl/thesis-tests.git
cd thesis-tests

2. Set up the virtual environment

python -m venv .venv
source .venv/bin/activate    # Linux/macOS
# .venv\Scripts\activate     # Windows

pip install -r requirements.txt

3. Configure API keys

cp .env.example .env
# Edit .env and fill in your API keys:
#   FRED_API_KEY=your_fred_key_here
#   BEA_API_KEY=your_bea_key_here

4. Download all data

This fetches data from all public APIs and builds compiled datasets. Takes 5-15 minutes on first run.

python download_data.py --source all

To verify data without re-downloading:

python download_data.py --verify-only

5. Run all tests

python run_all.py

Results are saved to results/scorecard.json.

Running individual tests

# Run a single test
python run_all.py --test test_aggregation_puzzle

# Run all tests in a folder
python run_all.py --folder foundations

# Print last scorecard
python run_all.py --scorecard-only

Data sources

The download script fetches from these public APIs:

• FRED — ~100 series (CPI, IP manufacturing, macro aggregates, PPI, employment)
• BEA — Regional GDP/compensation/GOS/employment, NIPA tables, I-O tables, Fixed Assets
• World Bank — WDI indicators (GDP, trade, Gini, financial depth) and WGI governance
• OECD STAN — Structural analysis database (25 sectors, 38 countries)
• Census BDS — Business Dynamics Statistics (firm entry/exit by sector)
• NBER-CES — Manufacturing industry database

A small number of hand-compiled datasets (literature values, academic paper tables) are committed in data/compiled/.

Test structure

Tests are organized into 8 folders mirroring the Lean formalization: