mksc-methods

Methods & Identification (mksc-methods)

When to trigger

• You must choose between a structural, analytical, or reduced-form/causal-ML approach
• The model is written but not yet estimable (parameters, moments, normalization)
• Identification is hand-waved ("we use instruments") without specifics
• A reviewer says "the design cannot identify the structural parameters"

Choose the genre that fits the claim

Marketing Science is methodologically plural around a modeling core: structural econometrics, analytical models, econometric/statistical analysis, ML tools, surveys, and experiments — all judged by whether they develop, test, or rigorously apply a formal model.

| Claim / goal | Approach that earns it |

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

| Quantify demand and simulate a policy | Structural demand (BLP/mixed logit), supply FOCs, counterfactual |

| Forward-looking behavior, adoption, churn | Dynamic discrete choice / dynamic games (Rust, BBL, CCP) |

| Strategic-interaction insight, comparative statics | Analytical (game-theoretic) model |

| Bidding, sponsored search, marketplaces | Auction/structural-IO model with equilibrium bidding |

| Heterogeneous treatment effects tied to a model| Causal ML (double/debiased ML, causal forests) disciplined by theory |

| Causal effect from field variation | Field experiment / quasi-experiment as identifying variation |

A field experiment or quasi-experiment is welcome when it identifies a model primitive or validates a mechanism, not as a stand-alone reduced-form result.

Make the model estimable and identified

• Estimator: match it to the model — GMM (BLP moment conditions), MLE/SMLE, simulated method of moments, or hierarchical Bayes (MCMC) for rich heterogeneity.
• Instruments / identifying variation: name them concretely (cost shifters, BLP/Hausman/differentiation instruments, exclusion restrictions, randomized or discontinuity variation) and defend exogeneity.
• Normalizations and functional form: state outside-good normalization, scale/location normalizations, and which assumptions are substantive vs. convenience.
• Computation: specify the solver, equilibrium/inner-loop fixed point, starting values, and how you handle multiple equilibria or local optima.

For analytical papers

Specify the equilibrium concept, solve it, and prove the claims; relegate long proofs to an appendix but state the key steps. Plan to validate counterintuitive predictions and discuss robustness to the modeling assumptions that drive them.

Execution bridge (StatsPAI / Stata MCP)

For the empirical / causal lane, estimate and audit rather than only specify. Full

map: [execution-with-mcp](../../../shared-resources/empirical-methods/execution-with-mcp.md). Marketing Science is heavily structural/analytical; the chain below serves its reduced-form / field-experiment lane — structural demand and analytical modeling are outside this causal-inference toolchain.

• detect_design → recommend → fit with as_handle=true → audit_result to

enumerate the checks the design owes.

• Panel / 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 and romano_wolf for the many-outcome

family-wise correction reviewers expect.

Match the toolchain to the reviewer pool, and report the effect size the venue

wants. 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

• [ ] Genre (structural / analytical / causal-ML / experiment) matches the claim
• [ ] Estimator matches the model (GMM/MLE/SMM/Bayes) and is stated
• [ ] Identifying variation/instruments named and exogeneity defended
• [ ] Normalizations and substantive vs. convenience assumptions separated
• [ ] Computation (solver, fixed point, starting values, multiple equilibria) planned
• [ ] Any experiment/quasi-experiment tied to a model primitive, not free-standing

Anti-patterns

• A structural model "estimated" with no stated moments or instruments.
• Reduced-form regressions presented as the whole contribution at MKSC.
• Ignoring multiplicity of equilibria or optimizer convergence.
• Treating convenience assumptions as if they were innocuous.

Methods pass for Marketing Science

Use this as a second-pass capability check. First lock the demand/supply mechanism, fit evidence, and counterfactual decision margin; then test whether the manuscript addresses quantitative marketing reviewers who read the model through the managerial counterfactual it makes possible.

• Primary move: Name assumptions, diagnostics, robustness, falsification, and failure modes; do not accept a method section that hides the decisive validity threat.
• Decision ledger: return claim / evidence / blocker / next edit rows so the next pass can patch the manuscript directly.
• Neighbor test: compare against Journal of Marketing Research for empirical marketing breadth, Management Science for wider OR/MS reach, Quantitative Marketing and Economics for specialist modeling; if the neighboring outlet has the stronger audience claim, recommend re-routing before polishing.
• Verification floor: before submission-ready advice, re-open resources/official-source-map.md for volatile rules and name the one unresolved fact that could change the recommendation.

Output format

【Genre】structural / analytical / causal-ML / experiment
【Model→estimator】GMM / MLE-SMLE / SMM / hierarchical Bayes
【Identification】instruments/variation → parameters; exogeneity defense
【Normalizations/assumptions】substantive vs. convenience
【Computation】solver, fixed point, starting values, multiplicity
【Next step】mksc-data-analysis