nhandyjr/Covert_Intervention_IQ

Estimating the causal effects of CIA‑involved coups on institutional quality in Sub‑Saharan Africa using declassified JFK, RFK, and MLK records. Combines panel IV, causal forests, and synthetic control methods # # Intelligence-Inteventions and Institutional Quality in Sub‑Saharan Africa **A Data Science Pipeline for Causal Inference using Declassified Records (JFK, RFK, MLK Files) + Panel Data** This repository contains the full data collection, processing, and causal analysis pipeline for my research project: > *Estimating the causal effect of CIA‑involved military coups on long‑term institutional quality (WGI Rule of Law, Control of Corruption, Government Effectiveness) in Sub‑Saharan African countries.* The project uses a novel instrument – **historical CIA base presence (early 1960s)** – extracted from the 2025 declassified JFK, RFK, and MLK assassination records via web scraping and named‑entity recognition (NER). Causal effects are estimated using panel IV, causal forests, and synthetic control methods. --- ## 🧠 Research Questions 1. **Average effect** – What is the causal effect of any military coup on subsequent institutional quality? 2. **Heterogeneity by foreign involvement** – Do coups with documented CIA involvement have systematically different effects? 3. **Moderation by colonial heritage** – Does the effect differ between extractive vs. settler colonies (Acemoglu, Johnson & Robinson 2001)? --- ## 📦 Repository Contents | File / Folder | Description | |---------------|-------------| | `data/` | Will contain raw and processed data (WGI, CSP coup events, colonial heritage, CIA involvement indicators) | | `scripts/` | Python scripts for downloading, scraping, filtering, and NER | | `results/` | Output CSVs, NER entity lists, and model results | | `README.md` | This file | | `requirements.txt` | Python dependencies | | `jfk_complete_pipeline.py` | Main script to download static releases (2017‑2018, 2022) and scrape dynamic 2025 tables (JFK, RFK, MLK) using Playwright | | `filter_and_ner.py` | Filter documents by target African countries and CIA keywords; download matching PDFs; run spaCy NER | | `causal_models.py` | Panel fixed effects, 2SLS with CIA ba …