Theory and Model Craft (ecopol-theory-model)

When to trigger

• The empirical results need a model to map estimates into a welfare or policy-cost statement
• A quantitative/structural model runs the counterfactual policy but its mechanism is opaque to a non-specialist
• The model is more elaborate than the policy question requires (a discussant will ask "why all this machinery?")
• The counterfactual policy scenario is computed but the policymaker cannot see which assumption drives the headline number
• Reviewers/discussants want the welfare interpretation, not just reduced-form effects

EP's model standard: minimal, transparent, policy-bearing

EP is not a theory journal. A model earns its place only if it does policy work: it converts estimates into welfare, fiscal cost, or distributional statements; it disciplines a counterfactual policy; or it clarifies a mechanism that the reduced form cannot. The house style is accessible, so the model in the main text should be the smallest model that carries the policy conclusion, with derivations and extensions in the technical appendix. Two tests every EP model must pass:

1. The transparency test. Can a policymaker see which assumption produces the headline number, and how the number moves if that assumption changes? If the model is a black box, the policy discussant will not trust the recommendation.
2. The necessity test. Remove the model — does the policy claim survive? If yes, the model is decoration; cut it to the appendix. If no, the model is the contribution and must be defended.

Decision table: how much model does this paper need?

Craft moves

• Lead with the mechanism in words, then formalize. The policy reader should grasp the channel before seeing a single equation.
• Expose the key elasticity / sufficient statistic. EP readers want to know "the answer depends mostly on parameter θ, which we estimate / take from the literature as X." Make θ visible and its source explicit.
• Show the counterfactual's sensitivity to the one or two assumptions that matter, in the main text — this is what a policymaker actually needs.
• Calibration discipline: every calibrated parameter gets a source and a sensitivity range; "standard values" without citation invites a discussant takedown.
• Welfare framing: state whose welfare, under what social objective, and what is omitted (distributional incidence, GE feedback, political constraints).

Checklist

• The model passes the necessity test — without it the policy claim would not hold
• The mechanism is stated in plain words before formalization
• The key elasticity / sufficient statistic is named, sourced, and its value justified
• Counterfactual policy sensitivity to the one or two driving assumptions is shown in the main text
• Every calibrated parameter has a source and a sensitivity range
• Welfare/cost claims state whose welfare, the objective, and what is omitted
• Heavy derivations are in the technical appendix, not crowding the main text

Anti-patterns

• A general-equilibrium apparatus where a one-line sufficient-statistic argument would do — over-modeling for a policy audience
• A black-box structural model whose headline number the reader cannot trace to an assumption
• "Standard calibration" with no citations and no sensitivity range
• A welfare statement that hides the distributional incidence a policymaker most cares about
• Model derivations dominating the main text and burying the policy message

Output format

【Journal】Economic Policy (EP)
【Skill】ecopol-theory-model
【Model role】framing / welfare-accounting / structural-counterfactual / mechanism
【Necessity test】does the policy claim need the model? Y/N
【Key parameter】θ = <value> (source) ; headline sensitivity to θ
【Welfare framing】whose welfare / objective / what is omitted
【Main-text vs appendix】what stays / what moves to appendix
【Next skill】ecopol-robustness