PrzemekSzkodon/heterogeneity-in-cash-transfers

We apply causal machine learning to the Haushofer and Shapiro (2016) 'THE SHORT-TERM IMPACT OF UNCONDITIONAL CASH TRANSFERS TO THE POOR: EXPERIMENTAL EVIDENCE FROM KENYA' study # D300 Project — Haushofer & Shapiro (2016) ## Data Download UCT_FINAL_CLEAN.dta from Harvard Dataverse: dataverse.harvard.edu Place in a folder called dataverse_files/ in the same directory as the notebook. ## Environment conda env create -f environment.yml conda activate causal_ml ## Running Run all cells in main.ipynb in order. Figures save to output/ automatically. ## Structure 1. Data preparation 2. Randomisation validation 3. Baseline balance 4. ANCOVA benchmark ATE 5. Post-selection ATE estimation 6. Honest causal forest 7. BLP, GATEs and CLAN 8. OLS of tau_hat on X (not included in D300 report) 9. Policy trees (not included in D300 report)