rje-identification-strategy

Identification Strategy (rje-identification-strategy)

When to trigger

• Your structural model's identification (instruments, functional form, moments) is unargued
• A merger/regulation evaluation rests on TWFE over staggered timing with no modern estimator
• An IV's first stage or exclusion restriction is weak in an IO setting
• You are unsure your design clears the RJE industrial-organization bar

The RJE identification bar

RJE is the flagship industrial-organization journal, so identification is judged by IO norms: the economic model and the source of identifying variation must be explicit, and counterfactuals must be disciplined by the model and the data. Both structural and reduced-form work is welcomed — there is no preference for one, but each carries field-specific obligations.

Branch A: Structural demand (BLP-style)

• Price endogeneity is the central threat. Instrument price with cost shifters, rival/own product characteristics (BLP instruments), or Gandhi–Houde differentiation instruments; argue their validity in market terms.
• Add micro-moments (consumer-level data) where available to pin down substitution and heterogeneity.
• State functional-form assumptions (random-coefficients distribution) and show key elasticities are reasonable, not artifacts.

Branch B: Supply, conduct & markups

• Be explicit about the conduct assumption (Nash-Bertrand, Cournot, collusion) and, where possible, test conduct rather than assume it.
• Identify marginal cost from the supply-side FOCs; show pass-through and markups are economically sensible.

Branch C: Entry / dynamic games

• For static entry, address multiplicity of equilibria (bounds, equilibrium selection).
• For dynamic models, justify the CCP / two-step (Hotz–Miller) approach or full-solution estimation, and state the state space and discount factor handling.

Branch D: Auctions

• Map the observed bids to the value distribution through the equilibrium first-order conditions; state the information paradigm (private vs common values).

Branch E: Reduced-form (mergers / regulation)

• Name the policy or merger shock giving exogenous variation; defend it institutionally.
• Staggered timing? Move beyond TWFE — use Callaway–Sant'Anna, Sun–Abraham, or de Chaisemartin–D'Haultfœuille, with an event-study plot and a Goodman-Bacon check.
• For IV: strong first stage, weak-IV-robust inference, and an exclusion restriction argued from market institutions.

Execution bridge (StatsPAI / Stata MCP)

Estimate and audit the design, don't only describe it. Full map:

[execution-with-mcp](../../../shared-resources/empirical-methods/execution-with-mcp.md). RAND is industrial organization — endogeneity of prices/entry and structural demand; the reduced-form chain for causal claims, structural IO outside it.

• 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](../../../Journal-of-Finance-Skills/resources/worked-examples/02-execution-walkthrough.md).

Checklist

• [ ] Economic model / identifying variation stated in one sentence
• [ ] Price endogeneity (structural) or treatment exogeneity (reduced-form) defended
• [ ] Instruments named and their validity argued in market terms
• [ ] Conduct / equilibrium-selection assumptions made explicit
• [ ] Modern estimator used where TWFE would be biased
• [ ] Counterfactual assumptions bounded and stated
• [ ] Claims never exceed what the model + data support

Variation-to-parameter map (what pins down which primitive)

RJE referees want an explicit account of which variation identifies which structural object.

| Structural object | Identifying variation | Threat if absent |

|---|---|---|

| Own-price elasticity | Cost/input-price shifters orthogonal to demand shocks | Endogeneity biases elasticity to zero |

| Substitution / random coefficients | Rival characteristics across markets; micro-moments | Restrictive IIA, wrong merger effects |

| Marginal cost | Supply-side FOCs given conduct and demand | Cost confounded with markup |

| Conduct parameter | Demand rotations (slope- vs level-shifters) | Conduct assumed, not identified |

Worked vignette: identifying conduct in an entry model

Suppose you study airline entry on city-pair routes (illustratively 1,200 routes over six years, one potential entrant each) and ask whether incumbents deter entry.

• Identifying variation: endpoint market-size shifters move entry value; slot/gate cost shifters move fixed costs independently.
• Multiplicity: simultaneous entry admits multiple equilibria — use Ciliberto-Tamer-style bounds or an explicit selection rule, and report the parameter set.
• One-sentence map: "Demand shifters identify entry value, cost shifters identify fixed costs, and incumbent presence correlating with non-entry net of profitability identifies deterrence."

Referee-pushback patterns and the venue fix

• "Demand identification rests on functional form, not variation." Fix: show which excluded instruments move price versus substitution, and perturb the random-coefficient distribution to prove the elasticities are not artifacts.
• "The exclusion restriction is asserted, not defended." Fix: argue why the cost shifter is plausibly orthogonal to the demand shock here, and probe with an overidentification or placebo-instrument check.
• "Conduct is assumed Nash-Bertrand without test." Fix: run a conduct test (Rivers-Vuong or a markup restriction) or report results under competing conduct models.
• "Pre-trends violate the reduced-form design." Fix: show event-study leads, use the heterogeneity-robust estimator, and report the Goodman-Bacon decomposition.

Anti-patterns

• A demand system with price treated as exogenous, or weak/unargued price instruments
• Assuming Nash-Bertrand conduct when the data could test it
• TWFE on staggered merger/regulation timing with no heterogeneity-robust estimator
• Selling an in-sample fit as a credible far-out-of-sample counterfactual

Output format

【Approach】structural demand / conduct / entry-dynamics / auctions / reduced-form
【Identifying variation】one sentence
【Key assumptions】[instruments, conduct, equilibrium selection, exclusion]
【Diagnostics done】[elasticities, first-stage F, pre-trends, conduct test]
【Diagnostics missing】[...]
【Next step】rje-data-analysis