Topic Selection & Policy-Question Fit (aejpol-topic-selection)

When to trigger

• You have a clean empirical result but are unsure it is "an AEJ: Policy paper"
• The project could plausibly go to J. Public Economics, AEJ: Applied, or AER and you must pick
• A referee or colleague says the question is "narrow," "not a policy paper," or "so what?"
• You can state a finding but not yet a policy question with a welfare / cost-benefit / distributional stake

The AEJ: Policy bar — lead with the policy question

AEJ: Policy publishes the economic analysis OF policy: a paper is built around a policy question ("should this tax / mandate / subsidy / regulation exist, expand, or change, and at what welfare cost?") whose answer carries a welfare, cost-benefit, or distributional implication of broad interest to the AEA readership. Two halves must both be present from the first page:

1. A real policy lever. A specific instrument someone could pull — a credit, a tax schedule, an eligibility rule, an emissions standard, a transfer, a mandate, an enforcement regime. Not just "an interesting natural experiment."
2. A counterfactual / welfare reading. What changes, for whom, and is it worth it — a cost-benefit ratio, a marginal-value-of-public-funds (MVPF), an incidence/distributional split, or a calibrated welfare number. A clean estimate with no policy reading is off-fit.

Policy areas in scope

Public economics & taxation · environmental & energy · health · education · labor & social insurance · regulation & antitrust · development policy · political economy of policy. Empirical (quasi-experimental / RCT) and applied-theory work both fit — provided the policy question and welfare relevance are explicit.

Fit decision table (route by the dominant pull)

Checklist

• [ ] The policy lever is named in one sentence (instrument + who is affected)
• [ ] The policy question is stated as a question with a welfare/cost-benefit/distributional stake
• [ ] The counterfactual is concrete (what the policy is compared against)
• [ ] Broad-interest test passed: a non-specialist AEA reader sees why it matters
• [ ] Sibling check done (not JPubE field-only / not AEJ:Applied no-policy / not AER-scale)
• [ ] You can name the welfare object you will eventually report (MVPF, cost-per-X, incidence)

Anti-patterns

• "We exploit a clean natural experiment" with no policy the experiment evaluates (reads as AEJ: Applied)
• A field-public-finance result with no broad-interest framing (reads as J. Public Economics)
• A descriptive or correlational "policy-relevant" topic with no credible counterfactual
• Promising a welfare/cost-benefit reading you have no way to compute
• Leading with the dataset or method instead of the policy question

Three questions that decide fit fast

Before investing in a draft, answer these in one sentence each; a "no" or "I can't" on any is a fit problem:

1. The lever test — can you name the instrument a decision-maker would pull? (If it is "a shock," not a policy, lean AEJ: Applied.)
2. The welfare test — can you name the welfare object you will report (MVPF, cost-per-X, incidence)? (If not, the policy "so what" is missing.)
3. The broad-interest test — would a non-specialist AEA reader, not just the field, care about the answer? (If only the field cares, lean JPubE.)

Worked vignette (illustrative)

A draft estimates that a state's expansion of a childcare subsidy raised maternal employment. As "we find subsidy → employment" it is a clean applied-micro result (AEJ: Applied). Reframed for AEJ: Policy: "Is expanding the childcare subsidy a cost-effective way to raise maternal labor supply, and who bears the cost?" — now the employment elasticity feeds a cost-per-additional-worker and an incidence split across income groups (illustrative), and the paper has a policy lever, a counterfactual, and a welfare reading.

Referee pushback mapped to the fix

• "Better suited to a field journal." → Sharpen the broad-interest framing; lead with the policy lesson, not the institutional detail.
• "This is just a clean natural experiment." → Name the policy the experiment evaluates and the welfare object; if there is none, reconsider the target.
• "Interesting but so what for policy?" → Add the cost-benefit / incidence reading to the abstract, not the conclusion.

Output format

【Policy lever】instrument + affected population (one sentence)
【Policy question】stated as a question with a welfare/cost-benefit/distributional stake
【Counterfactual】what the policy is compared against
【Welfare object to report】MVPF / cost-per-X / incidence / calibrated welfare
【Fit verdict】AEJ: Policy vs JPubE / AEJ:Applied / AER + one-line reason
【Next step】aejpol-literature-positioning