jfqa-workflow - SKILL.md Agent Skill

name: jfqa-workflow description: Use when routing a Journal of Financial and Quantitative Analysis (JFQA) manuscript — decides which jfqa-* skill to use next across topic selection, literature positioning, identification, data analysis, contribution framing, tables/figures, writing style, the code-sharing archive, review process, submission preflight, and R&R rebuttal. Start here when unsure what to do next.

JFQA Workflow Router (jfqa-workflow)

Use this skill to orient a manuscript aimed at the Journal of Financial and Quantitative Analysis (JFQA) — an empirical and quantitative finance journal published by Cambridge University Press for the University of Washington Foster School of Business. It prints less than 9% of the 1,000+ papers submitted yearly, so every stage should clear a high bar before you spend the $350 non-refundable-by-$75 submission fee.

When to trigger

"Where do I start / what's next for my JFQA paper?"
You have results but are unsure whether the design, framing, or archive is JFQA-ready.
You received a decision letter and need to route to rebuttal or resubmission.

Routing map

jfqa-topic-selection          → is this a JFQA-fit finance question?
        ▼
jfqa-literature-positioning   → stake the contribution vs. the finance frontier
        ▼
jfqa-identification-strategy  → credible design (sorts/FMB, panel FE, DID, IV, RDD, event study)
        ▼
jfqa-data-analysis            → estimation, clustering, robustness, economic magnitudes
        ▼
jfqa-contribution-framing     → one-paragraph ≤100-word abstract + intro contribution
        ▼
jfqa-tables-figures           → finance-standard exhibits with self-contained notes
        ▼
jfqa-writing-style            → polish for a finance audience
        ▼
jfqa-replication-and-data-policy → build the JFQA Dataverse code/data archive
        ▼
jfqa-review-process           → what double-anonymous JFQA review involves
        ▼
jfqa-submission               → Editorial Manager preflight (PDF, fee, formatting)
        ▼
jfqa-rebuttal                 → R&R response letter

Output format

【Stage】where the manuscript is now
【Next skill】single jfqa-* skill to invoke
【Why】one-line rationale tied to a JFQA constraint
【Watch-outs】fee/refund, 100-word abstract, anonymization, code policy, prior-rejection disclosure

Always confirm volatile specifics on the official pages (see resources/official-source-map.md).

Router diagnostics

Ask the manuscript owner for the current bottleneck:

Question/scope unclear or outside finance: jfqa-topic-selection.
Closest finance competitor missing: jfqa-literature-positioning.
Design credibility weak: jfqa-identification-strategy.
Results not reproducible or magnitudes missing: jfqa-data-analysis.
Abstract or first page too diffuse: jfqa-contribution-framing.
Exhibits unreadable without text: jfqa-tables-figures.
Archive, pseudo data, or exception timing unresolved: jfqa-replication-and-data-policy.

The route should always reduce one JFQA-specific screen risk: fit, 100-word abstract, anonymization, identification, economic magnitude, code-sharing, fee/prior-rejection disclosure, or exhibit readability.

Stage gates before spending the fee

Gate	Exit criterion	JFQA risk averted
G1 Scope	the question scores high on the topic-selection rubric	$75 retained on a desk reject, plus months
G2 Design	identifying variation stated in one defensible sentence	the endogeneity report that kills most empirical R&Rs
G3 Numbers	headline magnitude expressed in finance units, scaled	the "only statistically significant" complaint
G4 Package	`run_all` regenerates every exhibit; pseudo-data plan set	a scramble at acceptance under the Code Sharing Policy
G5 Preflight	≤ 100-word abstract, anonymized text-searchable PDF, disclosures drafted	formatting/anonymity strikes and the one-year-ban trap

Do not let a manuscript pass a later gate while an earlier one is open; JFQA's screens fire in roughly this order.

Worked routing call (illustrative)

A manuscript arrives with strong results but a 138-word abstract, firm-only clustered standard errors never checked against a two-way scheme, and no pseudo-data plan for its CRSP/Compustat extracts. Route in this order: jfqa-contribution-framing first (the abstract cap is a hard constraint and the cheapest fix), then jfqa-data-analysis (inference robustness changes table contents, so it must precede exhibit polish), then jfqa-replication-and-data-policy (pseudo data takes days, and any exception must ride on the initial submission), and only then jfqa-submission. The principle: hard journal constraints and anything that can change numbers come before taste-level polish.

Re-entry points after a decision

Desk reject → jfqa-topic-selection or jfqa-contribution-framing, depending on whether fit or framing failed.
Reject after review, planning a substantially modified return → rebuild through the full chain and add the prior-rejection disclosure at jfqa-submission.
R&R → jfqa-rebuttal, looping back into jfqa-identification-strategy or jfqa-data-analysis for each evidentiary demand.

What can run in parallel

The Dataverse archive build (jfqa-replication-and-data-policy) can proceed alongside writing-style polish; both consume the frozen results, not each other.
Exhibit notes (jfqa-tables-figures) and the abstract (jfqa-contribution-framing) share the same headline magnitudes — draft them in one sitting so the numbers cannot diverge.
Anonymization is not a stage; it is a property every stage must preserve from first draft to resubmission.