brycewang-stanford Organizing Framework — the Survey's Spine (jel-organizing-framework)
When to trigger
- The evidence matrix is built but the draft would read like a list of papers
- Sections are named after topics ("Empirical studies", "Theoretical work") rather than ideas
- A reader could not predict what comes next or why papers are grouped as they are
- You cannot state in one sentence the argument the survey makes about the field
Why the spine is the whole game at JEL
The single most-cited reason JEL surveys fail is that they are annotated bibliographies: paper-after-paper summaries with no organizing idea. A great JEL survey imposes a structure the field did not have — a taxonomy, a unifying framework, a sequence of questions, or a simple model — that makes scattered work legible. The framework is the contribution; the citations are the evidence. Choose the spine deliberately:
| Spine type | Organizes the field by | Best when |
|---|---|---|
| Taxonomy | mutually-exclusive categories of mechanism/approach | the field is fragmented into incommensurable camps |
| Unifying model | a simple framework whose parameters index the studies | results disagree because they estimate different parameters of one structure |
| Question sequence | a logical chain of sub-questions | the field has a natural "first we must know X, then Y" order |
| Chronological/paradigm | how thinking evolved and why | the evolution of ideas is itself the lesson |
| Methods spectrum | from least to most credible designs | the survey's value is appraising what is actually known |
Pick one primary spine; a second axis can be a within-section ordering, but a survey with two competing spines reads as two surveys.
The test of a good framework
- Exhaustive + exclusive (MECE-ish): every important paper has exactly one natural home, and the categories do not bleed into each other.
- Generative: the framework predicts where gaps are — empty cells are open questions, not omissions.
- Reconciling: apparent contradictions in the literature become explained (studies disagree because they sit in different cells / estimate different objects).
- Portable: a non-specialist can restate the spine after one read and use it to slot a new paper they encounter.
Stress-test by trying to place 5 hard cases (papers that resist categorization). If three of them have no home, the spine is wrong — redesign before drafting.
The framework is also what lets you be selective without being incomplete: once each cell is defined, confirmatory studies can be cited in clusters within their cell while the prose discusses only the cell-defining work. A survey without a spine cannot do this — it must either summarize everything (bloat) or omit silently (gaps). Design the spine before you decide what to foreground.
Checklist
- One primary spine chosen (taxonomy / model / question-sequence / paradigm / methods)
- The survey's one-sentence argument about the field is written
- Every category traces to evidence-matrix rows (no empty rhetorical buckets)
- The framework reconciles at least one apparent contradiction in the literature
- Empty cells are surfaced as open questions (generativity)
- 5 hard-case papers each have a natural home
- Section headings name ideas/mechanisms, not "theoretical vs. empirical"
- A non-specialist could restate the spine and slot a new paper into it
Anti-patterns
- The annotated bibliography: paper-by-paper summaries with no organizing idea (the cardinal JEL sin)
- Sections named "Theory", "Evidence", "Other" — categories that carry no analytical content
- A taxonomy whose categories overlap so every paper is cited three times in three places
- Two competing spines fighting for control of the same survey
- A framework so bespoke only the author can apply it (not portable)
- Hiding the contribution: never stating, in one sentence, what the survey argues about the field
Output format
【Spine type】taxonomy / unifying-model / question-sequence / paradigm / methods-spectrum
【Argument about the field】"<one sentence the survey makes>"
【Categories】<the cells / sub-questions, each MECE>
【Reconciliation】<which contradiction the framework explains>
【Open questions】<empty cells surfaced as gaps>
【Hard-case test】5 awkward papers each placed? Y/N
【Next step】→ jel-comprehensiveness-and-balance (fill cells fairly + even-handedly)