jbes-identification-strategy

name	jbes-identification-strategy
description	Use when the methodological core of a Journal of Business & Economic Statistics (JBES) paper is the bottleneck — assumptions, regularity conditions, asymptotic theory, and Monte Carlo design for a new estimator, test, or algorithm. Stress-tests the method's validity, not a causal design.

Method Validity & Asymptotics (jbes-identification-strategy)

When to trigger

• The estimator/test is proposed but its assumptions and limiting theory are not nailed down
• The asymptotic distribution, consistency, or rate is asserted without full conditions
• The Monte Carlo design does not yet show where the method holds and where it breaks
• For an applied paper, what parameter your method identifies (and under what conditions) is unclear

What "identification" means at a methods journal

At JBES the load-bearing question is usually not a causal-design story but whether the method delivers valid inference for its target under stated conditions. Because JBES demands methodological novelty with clear empirical relevance, the assumptions cannot be so strong that no real data set — including the paper's own application — satisfies them. The credibility ladder a referee applies, strongest first:

1. Identification of the target — the parameter/functional is identified from the population moments the method uses, under stated conditions.
2. Regularity conditions — explicit (moment existence, smoothness, mixing/dependence, rate and rank conditions), each motivated and as weak as the result allows.
3. Asymptotic theory — consistency, limiting distribution, convergence rate, and the asymptotic variance; uniformity if claimed.
4. Finite-sample evidence — Monte Carlo on size, power, coverage, bias/RMSE across DGPs and sample sizes, including assumption-stressing regimes.
5. Method robustness — behavior under dependence, heavy tails, weak identification, high dimension, or misspecification, as the problem invites.

Branch paths

• New estimator — identifying moment/objective; consistency and limiting distribution under explicit conditions; a consistent (often HAC/cluster-robust) variance estimator.
• New test — null, alternative, and null limiting distribution; asymptotic and finite-sample size control plus power against relevant alternatives; nuisance-parameter and boundary handling.
• Computational/algorithmic — correctness (same target as the slow method), complexity/scalability gain, numerical accuracy, timing, and the feasible-application payoff.
• Applied paper — what parameter is identified and the assumptions the method requires; dependence/weak-identification-robust inference; defend that the application needs the novelty.

Execution bridge (StatsPAI / Stata MCP)

Estimate and audit the identification claim, don't only argue it. Full map: execution-with-mcp. JBES is a business / economic-statistics venue — reviewers weigh estimator validity and simulation evidence, so pair every estimate with its diagnostics and, where relevant, a Monte-Carlo check.

1. detect_design → recommend → fit with as_handle=true → audit_result to list the checks the design still owes.
2. Staggered DiD: callaway_santanna / sun_abraham + bacon_decomposition + honest_did_from_result (the pre-trend test is low-power, Roth 2022).
3. IV: effective_f_test + an anderson_rubin_ci (valid under weak instruments), not a 2SLS t-stat alone.
4. RDD: rdrobust (bias-corrected) + rddensity / mccrary_test for manipulation.
5. OVB: oster_delta / sensemakr — how strong a confounder would have to be.

Report the economic magnitude; route the full battery to the appendix; keep every number reproducible. A run end-to-end (synthetic data, real returns) is in the JF execution walkthrough. If StatsPAI/Stata are not connected, adapt the vendored resources/code/ skeleton and flag any unverified number.

Checklist

• Target parameter/functional and its identification conditions stated
• Regularity conditions explicit and as weak as the result allows
• Consistency, limiting distribution, and rate established under those conditions
• A consistent, robust variance estimator provided
• Monte Carlo covers size, power, coverage, bias/RMSE across DGPs and n
• Breakdown regimes shown, not hidden; MC standard errors reported
• The required conditions are plausible in the empirical application

Anti-patterns

• Asserting an asymptotic distribution with no full set of conditions
• Assumptions so strong no real data set (or the paper's own application) satisfies them
• Monte Carlo only under favorable DGPs; no breakdown shown
• Rejection rates reported with no Monte Carlo standard errors
• A "robust" variance claim with no proof or coverage simulation

Worked vignette: nailing down a weak-IV-robust estimator's validity

A hypothetical JBES paper proposes a debiased estimator for a structural elasticity under many weak instruments, applied to demand estimation on scanner data (numbers illustrative). The credibility ladder forces order: (1) the target elasticity is identified from the conditional-moment restriction as instrument strength shrinks; (2) regularity conditions are weak — finite fourth moments and a concentration-parameter rate; (3) the limiting normal distribution and rate are established with a consistent, heteroskedasticity-robust variance; (4) Monte Carlo across a concentration-parameter grid shows coverage of an illustrative 93.8% near nominal 95% where 2SLS collapses; (5) the breakdown under heavy-tailed shocks is shown. Crucially, the scanner application's first stage is weak — exactly the regime the conditions target — so the assumptions are plausible for the paper's data.

Referee-pushback patterns on method validity (venue-specific fixes)

JBES referee objection	Fix this skill enforces
"Your assumptions hold for no real dataset, including yours."	Weaken conditions to what the result needs; show the application satisfies them
"The asymptotic distribution is asserted without full conditions."	List every regularity condition the limit theorem uses, each motivated
"The robust variance claim is unproven."	Prove consistency of the variance estimator and confirm coverage in simulation

Calibration anchor (hedged): at JBES, "identification" usually means valid inference for the target under stated conditions, not a causal-design narrative — but the conditions must be plausible for business/economic data, since a real application is part of scope. Where a condition's necessity is uncertain, state it as sufficient and flag the gap.

Output format

【Object】estimator / test / algorithm / applied-target
【Target + identification】parameter and conditions: ...
【Regularity conditions】[listed] — as weak as possible? [Y/N]
【Asymptotics】consistency / distribution / rate / variance estimator [status each]
【Monte Carlo】DGPs, n grid, size/power/coverage, MC SEs, breakdown shown? [Y/N each]
【Empirical plausibility】conditions hold in the application? [Y/N]
【Next step】jbes-data-analysis