Policy impact modeling

Summary

Rigorous evaluation and forward simulation of policy interventions. DiD, RDD, synthetic control, double ML, causal forest. Counterfactual simulation of policy bundles using a five-layer digital-twin methodology covering institutional graph, policy domains, legal frameworks, committees, and peer-country comparisons.

What we do

Policy impact modeling is the discipline of producing defensible causal claims about interventions that have happened, are happening, or are being designed. The work begins with a pre-analysis plan: identification strategy, parallel trends or RDD bandwidth justification, balance tests, and the placebo and robustness battery. Every estimation is paired with bounds (Manski, Rambachan-Roth) and the appropriate clustering or randomization-inference framing.

For forward-looking work, the consultancy uses a five-layer digital-twin methodology adapted from the Aegis platform: institutional graph, policy domains, legal frameworks, committees, and peer-country comparisons. The twin runs scenario bundles and propagates outcomes across policy areas in a single coherent simulation. Outputs are built for policy decision makers (memo, dashboard, scenario tree) and academic referees (replication package, identification appendix).

Methods

• Difference-in-differences, staggered DiD, synthetic DiD
• Regression discontinuity (sharp, fuzzy)
• Synthetic control with bounds (Rambachan-Roth, Manski)
• Double machine learning and causal forest
• Shift-share (Bartik) and randomization inference
• Event-study figures and placebo test batteries
• Five-layer digital-twin counterfactual simulation
• Pre-analysis plan and replication discipline

Public data partners

• BLS, BEA, FRED for US macro and labor
• World Bank WDI (via Sisyphus and Hercules warehouses)
• OECD, EU, IMF macro panels
• Sectoral administrative data as available
• OpenAlex and Semantic Scholar for literature integration

Deliverables

• Pre-analysis plan with identification strategy
• Full replication package (code, data, tables, figures)
• Policy brief suitable for non-technical decision makers
• Counterfactual scenario dashboard with sensitivity sliders
• Joint working paper draft if the engagement is co-authored

Sample engagements

• Causal evaluation of a national social transfer program with staggered DiD.
• Synthetic control for a state-level minimum wage change, 2019 to 2024.
• Five-layer digital-twin simulation of a tariff reform bundle for a developing economy.
• Pre-analysis plan and registration for a randomized health-financing pilot.
• Independent replication of a forthcoming working paper before public comment.