restat-workflow

REStat Workflow Router (restat-workflow)

Overview

This is the router. It tells you **which restat- skill to use at the current stage* of a manuscript aimed at The Review of Economics and Statistics (REStat) — the venerable MIT Press journal edited at the Harvard Kennedy School (founded 1917/1919; current title since 1948). REStat publishes applied economics and applied econometrics: it rewards credible causal identification, careful measurement, and quantitative evidence across labor, public, development, trade, IO, health, environment, urban, and macro-empirics. It is empirical-first — a clean research design or a measurement advance is the contribution; theory interprets, it does not lead.

Operational tells that you are at REStat and not a sibling: REStat values measurement and applied econometrics, not pure methods development (that is J. Econometrics); it is broad applied econ, wider than a field journal but more specialized than AER. Submission is via Editorial Express with a nonrefundable submission fee (US$125; source map refreshed 2026-06-20); review is editor-managed peer review; accepted empirical papers must deposit data and code to the REStat Harvard Dataverse under the journal's Data and Code Availability Policy. Editors as of the 2026-06-20 refresh: Will Dobbie (Harvard) and Raymond Fisman (Boston U), co-chairs.

When to trigger

• The user asks "what should I do next?"
• A draft needs its current bottleneck diagnosed
• Work is ping-ponging between identification, exhibits, framing, and the response letter
• A REStat decision letter arrived and the user needs to switch into revision mode

Routing table

| Current symptom | Next skill |

|-----------------|------------|

| Unsure the paper fits REStat vs AER / AEJ:Applied / J. Econometrics / a field journal | restat-topic-selection |

| Contribution vs. the closest prior work is fuzzy or undersold | restat-literature-positioning |

| Causal design / measurement strategy is the bottleneck | restat-identification |

| Theory is overgrown or absent; unclear how much model to carry | restat-theory-model |

| Headline estimate's stability under specs/sample/inference is untested | restat-robustness |

| Exhibits are dense; the main result is not legible in one table/figure | restat-tables-figures |

| The question + estimate do not land in the first paragraph | restat-writing-style |

| Need the Harvard Dataverse data/code package + README | restat-replication-package |

| Want to pre-empt the objections this design invites | restat-referee-strategy |

| Ready to submit via Editorial Express; need a preflight | restat-submission |

| Received an R&R; need a response-letter strategy | restat-rebuttal |

Default order

1. restat-topic-selection — lock the applied question and REStat fit
2. restat-literature-positioning — stake the marginal contribution
3. restat-identification — make the causal design / measurement credible
4. restat-theory-model — right-size theory to interpret the estimate
5. restat-robustness — show the headline survives specs, sample, inference
6. restat-tables-figures — make the result legible in one exhibit
7. restat-writing-style — land question + estimate early (intro last)
8. restat-replication-package — assemble the Harvard Dataverse package
9. restat-referee-strategy — pre-empt design-specific objections
10. restat-submission — Editorial Express preflight
11. restat-rebuttal — after the R&R

> restat-writing-style is a late-stage polish; do not rewrite the intro before identification and robustness settle.

Anti-patterns

• Deferring restat-replication-package until acceptance — REStat's Data and Code Availability Policy and Harvard Dataverse deposit are part of the path to publication, not an afterthought
• Letting restat-tables-figures polish exhibits while the identification is still moving
• Treating REStat like J. Econometrics (a new estimator is the contribution) or like AER (general-interest, agenda-setting scope) — REStat is broad applied econ with a measurement tradition

Routing by paper archetype

REStat spans several applied archetypes, and the bottleneck differs. Read the archetype, then enter the chain at the right link.

| Archetype | Likely first bottleneck | Enter at |

|-----------|-------------------------|----------|

| applied-micro causal design (DID / RD / IV) | heterogeneity-robust estimator + clean design | restat-identification |

| measurement / new data or index | construct validity + measurement-error discipline | restat-identification → restat-robustness |

| trade / IO / spatial (shift-share) | exposure-design inference + exogeneity of shares/shocks | restat-identification |

| structural-light applied | how much model to carry without becoming method-first | restat-theory-model |

Worked routing example (illustrative)

A user says: "My DID on a staggered state policy estimates fine, but a referee says TWFE is biased and the measure of my outcome is noisy." That is two distinct REStat pushbacks — heterogeneity-biased estimator (owned by restat-identification, Branch DID) and measurement error in the outcome (a REStat-signature concern, addressed in restat-identification measurement + restat-robustness). Route to restat-identification first; only once the headline effect is stable under Callaway–Sant'Anna and an attenuation correction do you return to restat-tables-figures and restat-rebuttal.

Minimal decision snippet

if decision_letter_arrived:        -> restat-rebuttal
elif ready_to_submit:              -> restat-submission
elif anticipating_referees:        -> restat-referee-strategy
elif need_data_code_package:       -> restat-replication-package
elif exhibits_dense:               -> restat-tables-figures
elif headline_not_robust:          -> restat-robustness
elif theory_mis-sized:             -> restat-theory-model
elif identification_or_measurement_shaky: -> restat-identification
elif contribution_fuzzy:           -> restat-literature-positioning
else:                              -> restat-topic-selection