Jue Replication Package

Replication Package (jue-replication-package)

When to trigger

• The paper contains empirical, simulation, or experimental work — JUE's mandatory replication policy applies
• The data is geocoded, restricted, or proprietary (addresses, parcels, admin records) and you need a confidentiality path
• You are at acceptance and must upload the replication folder before publication
• A spatial pipeline (geocoding, GIS joins, distance/ring construction) is undocumented and not reproducible
• A coauthor assumes the deposit can wait until after publication — it cannot

JUE's mandatory replication policy (检索于 2026-06；以官网为准)

JUE will publish a paper only if the data and code are clearly documented and readily available for replication. At acceptance, before publication, authors of papers with empirical work, simulations, or experiments must upload a replication folder to a major data repository. The minimum deposit is:

1. The dataset(s) used to generate all estimates reported in the paper;
2. A description of how intermediate datasets and programs were used to build the final dataset(s);
3. The complete set of code from raw data to final results — except where code would reveal confidential information about the data.

Proprietary / confidential data: authors may request an exemption from providing data and/or code. The request must be made in the submission cover letter to the editor. Even with an exemption, document the pipeline and provide whatever code does not reveal confidential information so the path to results is auditable.

Spatial-data deposits are harder than they look

Urban papers carry data hazards that generic replication checklists miss:

• Geocoding is not reproducible by default. Pin the geocoder, version, and date; deposit the crosswalk from addresses to coordinates (or a de-identified version) so distance/boundary variables can be regenerated.
• Restricted address/parcel data. Provide a synthetic or aggregated extract plus exact instructions to obtain the restricted file (agency, application, access tier), and deposit all code that runs against it.
• GIS layers and shapefiles. Deposit or cite the exact boundary/shapefile vintages (tracts, school zones, transit lines); boundaries change across years and silently break replication.
• Spatial construction code. Ring/donut buffers, distance-to-boundary, market-access, and spatial-weight matrices must be in the code, with the distance cutoffs and projection (CRS) stated.

Building the folder

1. One-click master script (run_all) raw → cleaned → estimates → exhibits, with a stated software/version environment.
2. README mapping each table/figure to the script that produces it; list data sources with access tiers and the GIS-layer vintages.
3. Separate raw, intermediate, and final data so reviewers can re-run from any stage.
4. Confidential path: if exempt, the README states what is withheld, why, and how a researcher with access can reproduce it; the cover-letter exemption is filed.
5. Verify on a clean machine before deposit — the most common failure is hard-coded local paths and unstated CRS/geocoder versions.

Restricted-data access paths common in urban work

Urban papers lean on data that cannot simply be posted; document the path, not just the absence:

• Census/admin microdata in an RDC/FSRDC enclave — state the dataset, the project-approval route, and deposit all code that runs inside the enclave plus disclosure-cleared output; a researcher with clearance must be able to reproduce.
• Proprietary real-estate / transaction data (CoreLogic, Zillow ZTRAX-type, MLS) — cite the vendor and vintage, deposit construction code, and provide a synthetic or aggregated extract where the license permits.
• Geocoded individual records — deposit the de-identified crosswalk or the construction code that maps protected addresses to the spatial variables (distance, ring, boundary side), withholding only the raw identifiers.
• Mobility / GPS / cell-phone data — document the provider, the spatial and temporal resolution, and the aggregation that produces the analysis file.

In every case the cover-letter exemption names what is withheld and why, and the README states exactly how an authorized researcher reproduces the results.