Theory & Mechanism Model (ecj-theory-model)
When to trigger
- A reduced-form result has no model or stated mechanism behind it ("what is the economics?")
- A structural/theory paper's model exists but its assumptions or comparative statics are not disciplined
- The empirical specification is not derived from, or connected to, any economic framework
- A referee will ask which economic force drives the result and you cannot point to one
Why this matters at EJ
EJ publishes the full breadth of economics — theory and applied — and both are read for whether the economic argument is sound and of broad interest. A clean empirical effect with no mechanism is a half-paper here: EJ wants to know why, not only whether, because the "why" is what travels across fields and gives the result general relevance. The model need not be elaborate; a one- or two-equation framework that delivers a sharp, testable comparative static, explained so a generalist follows it, is often stronger than a baroque one. EJ's exposition premium applies to theory too — the economic intuition is stated in words before the algebra. If the contribution is purely methodological/econometric with no substantive economics, weigh The Econometrics Journal (also RES) instead of EJ.
Two modes:
-
Theory-led paper: the model is the contribution; empirics (if any) illustrate or test it.
-
Empirics-led paper: a compact model or explicit mechanism organizes the empirics, generates the prediction being tested, and gives the estimate its interpretation.
Building / disciplining the model
Minimal-model discipline (for empirics-led papers)
- Write the smallest model whose comparative static is the empirical prediction. Tie one model object to one estimated coefficient.
- State agents, their objective, the constraint, and what they optimize over. The economic logic — substitution, incentives, market clearing — should be visible to a non-specialist.
- Derive the sign/shape prediction the data will test, and say what evidence would refute it.
- Check the partial-equilibrium result against general equilibrium: does the effect survive when prices/quantities adjust elsewhere? If not, say so and bound it.
Full-model rigor (for theory-led / structural papers)
- Assumptions: each one earns its place; flag which are substantive vs. technical convenience.
- Existence/uniqueness of equilibrium where relevant; comparative statics stated as propositions.
- Welfare analysis: who gains, who loses, and the efficiency benchmark.
- Map structural parameters to identifying variation before estimation (hand off to
ecj-identification).
- Robustness of qualitative conclusions to the key assumptions; counterfactuals.
Consistency checks a referee runs
- Do agents in the model behave optimally given prices and incentives?
- Does the market clear, and does the result survive equilibrium adjustment?
- Are there unmodeled selection or entry/exit margins that would overturn the conclusion?
- Is the comparative static the paper tests actually a prediction of the model, or assumed?
- Does the welfare claim follow from the model, or is it asserted?
- Could a generalist economist follow the economic intuition without the appendix?
Checklist
Anti-patterns
- An empirical result with no model and no mechanism — "we find an effect" with no economic story
- A model whose comparative statics are never connected to the data
- Partial-equilibrium conclusions that ignore obvious GE feedback a referee will raise
- A needlessly elaborate model that obscures rather than sharpens the economic point
- Algebra with no verbal economic interpretation — fails EJ's exposition bar
- Assumptions chosen for tractability that secretly drive the headline result
- Welfare statements that do not follow from the stated objective function
Output format
【Mode】theory-led / empirics-led
【Core mechanism】the economic force in one sentence
【Model primitives】agents / objective / constraint
【Testable prediction】comparative static + the estimated object it maps to
【Intuition-before-algebra】stated in words? [y/n]
【GE / selection check】survives equilibrium adjustment? [y/n + note]
【Welfare】who gains/loses; efficiency benchmark
【Next】ecj-identification (to test it) or ecj-robustness