Revacc Theory Development

Theory & Model Development (revacc-theory-development)

When to trigger

• Predictions read as bald associations ("disclosure X relates to outcome Y") with no friction
• You have archival results and are tempted to write predictions around them (HARKing risk)
• An analytical paper needs a clean information structure, equilibrium, and comparative statics
• A claimed channel (information asymmetry, agency, real effects) is asserted but not modeled
• A referee says "what is the friction?", "the model is not minimal", or "the comparative static has no accounting reading"

RAST is a genuine home for analytical accounting

Unlike outlets where modeling is marginal, RAST runs a real analytical lane alongside archival and experimental work. That raises two distinct bars depending on the lane, and the most common failure is mixing them up.

• Analytical / modeling. State primitives — players, information structure (who knows what, when), payoffs — then solve for equilibrium and present propositions with proofs and comparative statics. The RAST-specific demand: the model must be the minimal structure that delivers an accounting implication (optimal disclosure granularity, the demand for conservatism, signaling through reporting choices, audit-effort incentives, performance-measure design, the real effects of mandatory reporting). An elegant equilibrium with no accounting payoff is a math paper RAST will reject.
• Empirical archival. Derive directional predictions from accounting and information economics before estimation. Name the friction the accounting acts on and the channel verb (reduces information asymmetry, tightens monitoring, relaxes a covenant, disciplines investment, deters tax aggressiveness).
• Experimental / behavioral. Ground hypotheses in psychology plus accounting theory (judgment, motivated reasoning, professional skepticism, investor processing) where archival data cannot isolate the mechanism; the experiment manipulates the construct to establish the channel directly.

The mechanism chain (archival/experimental)

1. Setting & construct — the accounting object (disclosure, accrual, audit, tax position) and the institutional setting that makes it bite.
2. Friction — the information or agency friction (asymmetric information, adverse selection, moral hazard, contracting/monitoring force, attention constraint).
3. Channel — the economic process linking construct to outcome (pricing, contracting, real investment, litigation, reputation, enforcement).
4. Direction & form — sign; monotone, non-monotone, or conditional.
5. Cross-section / boundary — where the effect strengthens or reverses, and why economically.

Building the analytical model (RAST-style)

• Make the information structure explicit and defend each assumption as load-bearing — referees probe whether a simpler model gives the same result.
• Prove propositions; never assert equilibrium properties. Put proofs in an appendix, intuition in the text.
• Translate comparative statics into empirical predictions or policy implications an accounting reader cares about; pair the model with a stylized empirical illustration where feasible — RAST values theory that speaks to data.
• Name the closest existing disclosure/agency model and state the marginal modeling move (a new friction, a relaxed assumption, an added stage).

Checklist

• Every prediction rests on a named friction and channel, not a mechanical correlation
• Archival predictions were fixed before estimation (no HARKing)
• Analytical claims are stated as proven propositions with comparative statics
• The model is defended as the minimal structure for the result
• Each comparative static has an explicit accounting reading (not just an economics one)
• At least one rival explanation is named and will be tested or modeled out
• The marginal move over the nearest prior model/theory is stated

Anti-patterns

• HARKing: writing predictions to match regression output after the fact.
• Model without payoff: a clean equilibrium whose comparative statics carry no accounting implication.
• Non-minimal model: assumptions a referee can strip without changing the result.
• Borrowed friction: invoking "information asymmetry" with no account of how it operates here.
• Mechanical prediction: a sign that follows from accounting identities rather than economics.
• Channel by proxy: asserting a real-effects channel the design cannot separate from pricing.

Output format

【Lane】archival / analytical / experimental
【Friction】information asymmetry / agency / contracting / attention ...
【Analytical only】primitives + information structure stated? propositions proven? minimal? yes/no
【Prediction P1 (focal)】construct → channel → outcome; direction/form
【Prediction P2 (channel/mediation)】...
【P3+ (cross-section/boundary)】... economic reason
【Accounting payoff】the implication a reporting reader cares about
【Rival explanation】named + how it will be ruled out
【Next skill】revacc-literature-positioning, then revacc-methods

Source: brycewang-stanford/Awesome-Journal-Skills → Review-of-Accounting-Studies-Skills/skills/revacc-theory-development/SKILL.md