scholarpeer-econ

name	scholarpeer-econ
description	Multi-agent peer review simulation for finance/economics manuscripts. Generates 2-3 reviewer personas (econometrician, domain expert, methodologist), conducts independent reviews, then synthesizes through discussion phase. Use when: (1) Pre-submission quality check before journal submission, (2) Internal review of coauthor manuscripts, (3) Simulating journal review process, (4) Identifying weaknesses in working papers. TRIGGERS: 'scholarpeer', 'multi-review', 'panel review', 'simulate reviewers', 'pre-submission review'.
allowed-tools	["Read","Write","Edit","Bash","Agent","Glob","Grep"]

ScholarPeer-Econ: Multi-Agent Peer Review for Finance and Economics

Overview

ScholarPeer-Econ simulates a journal peer review panel by dispatching multiple specialized reviewer agents in parallel. Each reviewer evaluates the manuscript from their expertise perspective, then a discussion phase synthesizes findings into a consensus report with actionable recommendations.

Based on the ScholarPeer framework, adapted for finance and economics research conventions.

When to Use This Skill

• Pre-submission quality check before sending to journal
• Internal review of working papers or coauthor drafts
• Identifying weaknesses that real referees would flag
• Simulating RFS, JF, JFE, or AER review process

Workflow

Phase 1: Setup and Persona Generation

1. Read manuscript at the provided path
2. Identify journal target (if specified) or use default balanced criteria
3. Select reviewer personas based on paper type:

Paper Type	Reviewer Configuration
Empirical corporate finance	Econometrician + Domain Expert + Methodologist
Empirical asset pricing	Econometrician + Domain Expert + Methodologist
Theoretical	Domain Expert (theory focus) + Methodologist
Experimental/survey	Methodologist + Domain Expert

4. Load journal profile from references/journal-profiles.md if target specified

Phase 2: Independent Review (Parallel Dispatch)

Dispatch 2-3 reviewer agents in parallel using the Agent tool. Each reviewer receives:

• Manuscript content
• Persona-specific evaluation criteria (see references/reviewer-personas.md)
• Scoring rubric (see references/scoring-rubric.md)

Dispatch pattern:

Single message with multiple Agent tool calls:
- Agent 1: Econometrician review
- Agent 2: Domain Expert review
- Agent 3: Methodologist review

Each reviewer produces:

• Individual scores on 5 dimensions
• Major comments (numbered)
• Minor comments (numbered)
• Specific recommendations

Phase 3: Discussion and Consensus

After all reviews complete:

1. Identify disagreements: Dimensions where reviewers differ by >1 point
2. Facilitate discussion: Present differing views, ask reviewers to justify positions
3. Resolve conflicts: Either reach consensus or document the disagreement
4. Synthesize: Merge into unified assessment

Use scripts/synthesize_reviews.py to aggregate scores and identify disagreements.

Phase 4: Consensus Report Generation

Create the final consensus report at:

quality_reports/reviews/YYYY-MM-DD_[paper-name]_consensus-report.md

Report structure:

# Peer Review Consensus Report

**Manuscript:** [Title]
**Journal Target:** [Journal or "General"]
**Review Date:** YYYY-MM-DD
**Reviewers:** Econometrician, Domain Expert, Methodologist

## Executive Summary
[2-3 sentences: overall assessment and recommendation]

## Consensus Scores

| Dimension | Score | Confidence |
|-----------|-------|------------|
| Identification Strategy | X/5 | High/Medium/Low |
| Data Quality | X/5 | High/Medium/Low |
| Robustness | X/5 | High/Medium/Low |
| Contribution | X/5 | High/Medium/Low |
| Clarity | X/5 | High/Medium/Low |
| **Overall** | **X/100** | — |

## Major Issues (Must Address)
1. [Issue with specific recommendations]
2. [...]

## Minor Issues (Should Address)
1. [...]
2. [...]

## Strengths
- [Strength 1]
- [Strength 2]

## Recommendation
[Accept / Minor Revision / Major Revision / Reject]

## Reviewer Discussion Notes
[Summary of any disagreements and resolution]

Reviewer Personas

See references/reviewer-personas.md for detailed specifications.

Quick Reference

Persona	Subagent Type	Primary Focus	Can Dispatch
Econometrician	methods-referee	Identification, causality, specification	strategist-critic
Domain Expert	domain-referee	Literature, contribution, institutional context	librarian
Methodologist	methods-referee	Data quality, measurement, replication	coder-critic

Agent Integration

ScholarPeer-Econ integrates with the existing research pipeline:

Integration Points

Econometrician reviewer:
  - Can dispatch strategist-critic for identification audit
  - Checks DiD/IV/RDD assumptions against strategy memo

Domain Expert reviewer:
  - Can dispatch librarian for literature search
  - Checks positioning against frontier map

Methodologist reviewer:
  - Can dispatch coder-critic for data/code review
  - Checks replication package completeness

When to Dispatch Subagents

• Identification concerns: Econometrician dispatches strategist-critic
• Literature gaps: Domain Expert dispatches librarian
• Data/replication concerns: Methodologist dispatches coder-critic

Scoring Alignment

The 100-point overall score aligns with the existing quality gate system:

• >= 95: Ready for submission
• 90-94: Minor revisions needed
• 80-89: Major revisions needed
• < 80: Fundamental issues require rethinking

Journal Profiles

See references/journal-profiles.md for detailed expectations by journal.

Journal	Emphasis	Typical Threshold
RFS	Theoretical contribution, novelty	95+
JF	Identification rigor, causality	95+
JFE	Empirical contribution, data quality	92+
AER	General interest, methodology	95+

Output Files

All outputs saved to quality_reports/reviews/:

File	Content
YYYY-MM-DD_reviewer-1-econometrician.md	Individual review
YYYY-MM-DD_reviewer-2-domain-expert.md	Individual review
YYYY-MM-DD_reviewer-3-methodologist.md	Individual review
YYYY-MM-DD_discussion-transcript.md	Discussion record
YYYY-MM-DD_consensus-report.md	Final synthesized report

Critical Rules

1. Parallel dispatch: Always dispatch reviewers in a single message with multiple Agent calls
2. Independent first: Reviewers must not see each other's reviews during Phase 2
3. Document disagreements: If consensus cannot be reached, document both positions
4. Actionable recommendations: Every major issue must include specific recommendations
5. Score calibration: Use the rubric in references/scoring-rubric.md for consistency

Example Usage

/scholarpeer-econ paper/main.tex --journal JF

This will:

1. Read the manuscript at paper/main.tex
2. Load JF journal profile (identification emphasis)
3. Dispatch 3 reviewers in parallel
4. Synthesize into consensus report
5. Save to quality_reports/reviews/2026-04-18_main_consensus-report.md

References

• reviewer-personas.md — Detailed persona specifications
• evaluation-criteria.md — Finance/econ-specific criteria
• scoring-rubric.md — Score definitions and calibration
• journal-profiles.md — RFS, JF, JFE, AER expectations