aejpol-theory-model

Theory / Welfare Model — Estimate-to-Policy Bridge (aejpol-theory-model)

When to trigger

• You have a credible causal estimate but no framework to say what it means for welfare or policy
• A referee says the welfare claim is "hand-waved" or the "so what" is missing
• You need to convert an elasticity / treatment effect into a cost-benefit or optimal-policy statement
• A reviewer asks "what is the sufficient statistic, and does your estimate identify it?"

The AEJ: Policy role of theory

At AEJ: Policy theory is usually in service of the policy reading, not the headline. Most papers do not need a full structural model; they need a transparent framework that turns a reduced-form estimate into a welfare, cost-benefit, or distributional object a policymaker can use. Pick the lightest framework that delivers the policy statement and make its assumptions explicit.

Bridge paths (lightest first)

Path A: Sufficient statistics / MVPF

• Write the welfare expression and show which estimated objects are the sufficient statistics (e.g., an elasticity, a fiscal externality, a pass-through). State that your design identifies exactly those.
• For spending/tax policies, a Marginal Value of Public Funds (benefit to recipients per dollar of net government cost) is the canonical AEJ: Policy summary — define the numerator and denominator and which estimates feed each.
• State the assumptions the sufficient-statistic formula buys you (envelope conditions, no income effects, partial-equilibrium scope) and where they could fail.

Path B: Cost-benefit / fiscal accounting

• Build the explicit ledger: program cost, behavioral-response fiscal effects, benefits to recipients, externalities. Show the net cost per unit of outcome (cost per job, per ton abated, per QALY-equivalent, per child lifted) with uncertainty propagated from the estimate's SE.
• Distinguish mechanical from behavioral effects; the behavioral term is what your causal estimate supplies.