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

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)