brycewang-stanford JEL Writing Style — the Survey Voice (jel-writing-style)
When to trigger
- The framework and coverage are settled and it is time to make the survey read well
- A non-specialist economist would get lost or bored in the current draft
- The opening buries the field's importance under machinery or throat-clearing
- The abstract reads like a paper abstract ("we find…") rather than a survey abstract
The JEL voice: authoritative + accessible + long-form
A JEL survey is read by non-specialist economists — smart, but not in this subfield. The voice is that of a trusted, fair expert teaching the field to a colleague from another area. Four qualities define it:
- Authoritative. Confident judgments about what is known, what is contested, and how good the evidence is. The reader wants the author's appraisal, not neutrality-as-evasion. But authority is earned by comprehensiveness and balance, never asserted.
- Accessible. Technical machinery is explained with intuition first, notation second. A method is motivated by the question it answers before any formalism. Jargon is defined or avoided; an outsider should follow the argument.
- Long-form and signposted. Surveys are long; the reader needs a map. Strong section/subsection architecture that mirrors the organizing spine, plus periodic "where we are / where we go next" cues, so the reader never loses the thread.
- Synthesis prose, not summary prose. Sentences make papers talk to each other ("Whereas X finds…, the design in Y suggests…"), rather than marching paper-by-paper.
The survey opening arc
A JEL introduction differs from a research-paper intro. The arc:
frame the field (what it studies, why it matters to economists broadly) → why a synthesis is needed now (new evidence / method shift / policy salience) → the organizing question and the spine (the map the reader will be given) → what the survey concludes and the open questions (the payoff, stated up front) → roadmap (brief).
State the bottom line early. A survey that withholds its conclusions to a final section wastes the non-specialist's time and reads as indecisive.
The abstract
A survey abstract describes what the survey does for the reader, not a single empirical finding: the field, why a synthesis now, the organizing idea, and the headline state-of-knowledge takeaway. Avoid "we estimate / we find a coefficient of…"; that is a primary-paper abstract. (Confirm any length/format limits in the AEA JEL style guide via jel-submission.)
Length and the reader's stamina
JEL surveys are long, and length is a privilege, not a license. Every section must earn its pages by advancing the argument about the field; coverage for its own sake (the confirmatory tier of jel-comprehensiveness-and-balance) belongs in compact citation clusters, not in paragraph-length summaries. Use subsection summaries and a recurring return to the spine so a reader who puts the survey down can re-enter it without rereading.
Checklist
- Opening follows the arc: frame field → why now → organizing question + spine → conclusions + open questions → roadmap
- Bottom-line state of knowledge stated early, not withheld to the end
- Technical material introduced intuition-first, notation-second; jargon defined or avoided
- Section architecture mirrors the organizing spine; periodic signposting present
- Prose synthesizes (papers in dialogue), not summarizes (paper-by-paper)
- Author's appraisals are explicit and labelled as the author's read where contested
- Abstract describes the survey's contribution to the reader, not a single estimate
- A non-specialist economist could follow start to finish
Anti-patterns
- The annotated-bibliography drone: "Smith (2010) finds… Jones (2012) finds…" with no connective argument
- Opening with method/notation before the field's question and stakes
- Withholding the survey's conclusions until the final section
- Notation-first exposition that locks out the non-specialist (JEL's core reader)
- A primary-paper abstract ("we find a coefficient of 0.3") on a survey
- Neutrality-as-evasion: refusing to appraise because appraising feels risky
- Under-signposting a long manuscript so the reader loses the thread
Output format
【Opening arc】frame → why-now → question+spine → conclusions+open-questions → roadmap? Y/N
【Bottom line early】state of knowledge stated up front? Y/N
【Accessibility】intuition-first; jargon controlled; non-specialist can follow? Y/N
【Signposting】architecture mirrors spine; periodic cues? Y/N
【Synthesis prose】papers in dialogue, not listed? Y/N
【Abstract】survey-style (contribution to reader), not a single estimate? Y/N
【Next step】→ jel-classification-system (assign JEL codes) → jel-submission