Imfer Identification

Identification Strategy (imfer-identification)

When to trigger

• A cross-country causal claim rests on OLS-plus-controls or TWFE on staggered policy adoption
• A high-frequency policy-surprise series (monetary, FX intervention) needs its exclusion defended
• A crisis "event study" cannot separate the policy from the macro shock that triggered it
• A capital-control or program effect is plausibly endogenous to the crisis it is meant to address
• An open-economy model's parameters are estimated but it is unclear what in the data identifies them

The IMFER identification bar

IMFER referees read as both frontier econometricians and policy analysts, so the mapping from data to the policy-relevant object must be explicit and the object must be the one a policymaker cares about. International-macro data make this harder than a clean single-country RCT: small N of countries, endogenous policy adoption, global common shocks, and spillovers that violate SUTVA across borders. Name the variation, defend it against the macro confounder, and report inference that respects cross-country dependence.

Branch A: Cross-country panel

• Move beyond TWFE under staggered policy adoption: Callaway–Sant'Anna, Sun–Abraham, de Chaisemartin–D'Haultfœuille; show clean event-study leads.
• Global common shocks (the global financial cycle, US monetary policy) confound country panels — saturate with time effects, or use a shift-share / external-instrument design.
• Spillovers break SUTVA: a control imposed in one country affects its neighbors. State the interference assumption or model the network explicitly.
• Inference: cluster on country and allow cross-sectional dependence (Driscoll–Kraay); with few countries, wild-cluster bootstrap.

Branch B: High-frequency policy-surprise

• Build the surprise from a tight window around announcements (intraday futures/yields); show it is unpredictable from the prior information set.
• Defend the exclusion: the surprise moves the outcome only through the policy channel, not contemporaneous news. Use a poll-based or sign-restriction purge if needed.
• For spillover designs, identify the foreign transmission (US-shock-to-EM-flows) and rule out the domestic-news confounder.

Branch C: Crisis event study / narrative

• Define the event window and the counterfactual path; the trigger shock and the policy response are nearly simultaneous — argue what isolates the policy.
• Use narrative classification (IMF program dates, intervention episodes) with documented coding rules; report sensitivity to window length and event definition.
• Address selection into crisis/program: countries adopt programs precisely when fundamentals deteriorate.

Branch D: Open-economy structural / DSGE

• Name what identifies each parameter from a data moment (an impulse response, a comovement, an external-finance premium) — not "the estimator converged."
• Report a sensitivity matrix (which moment moves which parameter) and Monte Carlo recovery of known parameters.
• Argue policy-invariance for the counterfactual (Lucas critique), since IMFER counterfactuals are read as policy advice.

Execution bridge (StatsPAI / Stata MCP)

Estimate and audit the design, don't only describe it. Full map: execution-with-mcp. IMFER is international macro/finance; cross-country panels with confounded policy — emphasize identification and clustering.

• detect_design → recommend → fit with as_handle=true → audit_result.
• Observational causal claims: staggered DiD (callaway_santanna / sun_abraham + bacon_decomposition + honest_did_from_result); IV (effective_f_test + anderson_rubin_ci); RDD (rdrobust + mccrary_test).
• Experiments: randomization-based inference + romano_wolf for many-outcome control.
• Sensitivity: oster_delta / sensemakr for observational claims.

Report the magnitude in interpretable units; route the full battery to the appendix. A run end-to-end (synthetic data, real returns) is in the JF execution walkthrough.

Checklist

• Branch chosen; the data-to-policy-object mapping stated in one sentence
• Global common shocks / the global financial cycle addressed in cross-country designs
• Cross-border spillovers (SUTVA) acknowledged and handled, not assumed away
• Staggered-adoption bias handled with a modern estimator; clean event-study leads shown
• Policy-surprise series shown unpredictable; exclusion defended institutionally and empirically
• Selection into crisis/program confronted, not ignored
• Inference respects cross-sectional dependence and few-country issues; SEs not asterisks
• Structural: each parameter tied to an identifying moment; sensitivity + recovery shown