Methodology v1.0.0 · reviewed 2026-07-11

Public Asking-Rate Index Methodology

Within each evidence-qualified geography, the facility-weighted constant-quality change in recurring, publicly advertised online monthly rent among matched, currently available unit configurations observed at covered, identity-verified U.S. self-storage facilities. This document defines the evidence required before any numeric release may carry the Storage Units Near Me price-index name.

Current status: collection and validation. The connected database has no observed price records, so no numeric Storage Units Near Me index is published.

Primary formula

Facility-weighted matched-product change

I(t) = I(t−1) × exp[ Σ w(f) × meanᵤ( ln p(f,u,t) − ln p(f,u,t−1) ) ]

Here f is a facility, u is an exact normalized unit configuration, and facility weights sum to one. Publication includes the match rate, sample composition, weighting rule, and uncertainty interval.

Target quantity and scope

For each evidence-qualified geography, the primary estimand is the facility-weighted constant-quality change in recurring, publicly advertised online monthly rent among matched, currently available unit configurations observed at covered, identity-verified U.S. self-storage facilities. It is a covered-panel new-customer asking-rent measure—not a claim about every U.S. facility, achieved rent, existing-tenant rent, occupancy cost, transaction price, or a cost-of-living index.

Unit of observation: facility × exact unit configuration × UTC collection timestamp.
Nominal U.S. dollars per month; taxes and unknown fees are not silently imputed.
Climate-controlled, non-climate-controlled, indoor, drive-up, floor, and vehicle products remain distinct.
Promotional and recurring prices are separate series unless a fixed-horizon effective cost can be calculated.

Source acquisition and rights

Validation collection is limited to explicitly allowlisted public facility pages. The crawler identifies itself, evaluates robots rules for its user agent and the wildcard agent, honors crawl delay, and never enters authentication, account, lease, reservation, payment, or protected API paths.

The exact page URL, retrieval timestamp, robots decision, source domain, parser, and raw-record hash are retained.
JSON-LD Offer and OfferCatalog records are preferred because price, unit, and availability remain structurally linked.
Review anecdotes and loose dollar-pattern matches are never observations.
Robots permission does not establish content-use rights. Research ingestion and aggregate publication require a versioned per-source rights record with a terms URL and checksum, review date, reviewer, permitted evidence fields, facility-path rule, and explicit aggregate-use decision. Raw third-party records are not redistributed.
Rights and field-semantics approvals are invalid when future-dated, more than 90 days old at collection, or contradicted by terms, path, or structured-template drift.
A licensed structured feed may enter under the same observation and provenance contract. Published third-party market reports remain separately labeled comparison benchmarks unless their license expressly permits record-level research ingestion.

Validation, normalization, and deduplication

An eligible observation requires an exact positive USD monthly price tied to parseable dimensions and a public facility identity. Every exclusion is retained as data rather than disappearing from the denominator.

Normalize dimensions, square feet, climate status, access type, floor, availability, and operator.
Flag generic ranges, stale offers, missing dimensions, impossible values, and unavailable products.
Deduplicate source × facility × unit configuration × observation period by a deterministic hash.
When multiple sources expose the same unit, preserve each source but prevent duplicate copies from receiving extra index weight.

Primary matched-product estimator

The primary time index matches the same facility and normalized unit configuration in adjacent periods. Price relatives are calculated in log space, aggregated within facilities, and then combined with stable facility weights so a property publishing many unit variants cannot dominate the series.

Unit match: facility × dimensions × climate × access × floor/vehicle class.
Within-facility aggregation occurs before national, state, or city aggregation.
Adjacent-period relatives are chain-linked to a base index of 100.
Entry and exit of unmatched products affect coverage statistics but not the matched relative directly.

Composition-adjusted sensitivity model

A secondary log-price model estimates time effects while controlling for square footage, climate control, access, floor, drive-up or vehicle status, geography, operator/source, and facility effects where the panel permits. It is a sensitivity estimate, not a replacement for the matched-product series.

Continuous size uses a prespecified nonlinear basis rather than a single linear assumption.
Time coefficients are transformed back to index relatives with an explicit retransformation rule.
Uncertainty is clustered or block-bootstrapped by facility.
Material divergence from the primary index triggers a documented composition review.

Sparse geography and partial pooling

Direct state and city estimates publish only when their own sample and confidence interval qualify. Hierarchical partial pooling may appear as a separately labeled model-adjusted estimate, but a national fallback is never presented as a local observation.

Suppressed estimates roll up to the nearest evidence-qualified parent.
Operator concentration is computed on effective facility weights, not raw offer counts.
Geographic coverage, missingness, matched share, and interval width appear beside every estimate.
Direct and model-adjusted values are never mixed under one unlabeled column.

Uncertainty, outliers, and sensitivity

Every qualified release reports median and mean together, quartiles, sample counts, and a 95% facility-clustered confidence interval. No automatic numeric outlier rule is part of methodology version 1.0.0.

Invalid, non-positive, and non-USD values are excluded; genuine high or low observed values are retained when their identity and semantics qualify.
Any future numeric outlier rule or sensitivity treatment requires a named methodology version and a disclosed comparability assessment.
Bootstrap resampling occurs at facility level to preserve within-property dependence.
Seasonal adjustment is withheld until at least three full years of a stable monthly series exist.

Release, revision, and reproducibility

Monthly releases begin as preliminary, may revise for three subsequent releases, and become final after 90 days. Content-addressed resources, overwrite checks, external preservation, checksums, code versions, manifests, seeds, and a public revision table support later reproduction.

A correction creates a new version and revision record; frozen artifacts are never silently overwritten.
Latest-series URLs point to the current release while content-addressed, externally preserved release resources stabilize citations.
Major methodology changes receive a new version and a linked methodological break note.
A DOI is added only after an external repository deposit exists; no identifier is invented.

Initial publication gates

These are provisional starting thresholds. Interval precision and sensitivity results can still suppress an estimate that clears the raw count. They are not treated as validated until simulation and backtesting approve the exact geography/gate version.

No geography may publish until its gate version passes documented simulation, backtesting, and false-precision review on representative panel conditions.

Level	Facilities	Diversity	Matched observations	Maximum core missingness	Maximum operator weight	Maximum relative CI width
National	500	25 states; 10 operator groups	≥50% products; ≥50% facilities	20%	35%; unknown 0%	20%
State	30	3 operators; 5 counties/metros	≥100 matches; ≥50% products; ≥50% facilities	20%	50%; unknown 0%	30%
City	12	3 operator groups	≥50 matches; ≥50% products; ≥50% facilities	20%	60%; unknown 0%	35%

Separate published series

Recurring advertised web rent

The recurring monthly online asking rent for an available unit, excluding temporary first-month promotions and separately stated fees.

Promotional first-month rent

The explicitly stated first-month promotional rent, published separately and never mixed with recurring rent.

Effective 3-, 6-, and 12-month rent

A fixed-horizon monthly equivalent calculated only when promotion duration and known mandatory charges are observable.

BLS gross-rent benchmark

External validation series WPU431104; a producer gross-rent index with a different estimand from new-customer online asking rent.

Pre-specified qualification exclusions

Robots or source policy does not permit collection of the public facility page.
The observation requires authentication, a lease/reservation flow, a protected API, or circumvention.
Currency is not USD or the price is missing, zero, or negative.
Exact unit dimensions cannot be parsed or the price cannot be tied to one unit configuration.
The quote is stale, unavailable, a review anecdote, a generic price range, or an unverified loose-text match.
A duplicate source/facility/unit observation has already been retained for the collection timestamp.
Core quality-adjustment attributes exceed the release missingness threshold.

Standards and external benchmarks

BLS Producer Price Index calculation methodology →FRED WPU431104 — Self-Service Storage and Miniwarehouses, Gross Rents →Google Dataset structured-data documentation →W3C Data on the Web Best Practices →W3C DCAT 3 →

Research status Data dictionary Quality gates