How Many Major U.S. Festivals Survive Ten Years?

Survival statistics for music festivals circulate widely and none of them has a published method. The obstacle is not analytical difficulty; survival analysis is standard. The obstacle is that no register of United States music festivals exists, so there is no cohort to follow and no denominator to divide by.

This document publishes the design rather than a result. A rate computed on an undefined sample is worse than no rate, because it acquires citations and outlives the caveat attached to it.

Defining the sample

A cohort requires four inclusion criteria fixed in advance. Scale: a minimum capacity or attendance threshold, since a 2,000-capacity one-day event and a 70,000-capacity three-day event are different businesses. Format: multi-artist, multi-hour, ticketed, recurring by design — excluding one-off concerts and free civic events, which have different failure conditions. Geography: United States, recorded by state and metropolitan area. Vintage: the first edition year, which defines cohort membership.

Each criterion excludes events that reasonable people would include, which is unavoidable. What matters is that they are stated, applied consistently and published with the result, so a reader can reconstruct the sample.

Defining survival

Survival is not a single condition. At minimum four states must be distinguished: continuous operation, meaning an edition took place in each scheduled cycle; interruption with return, where one or more cycles were skipped and the event resumed; transformation, where the event continues under a materially different name, site, format or owner; and cessation, where no further edition occurred. Treating transformation as survival produces a substantially higher rate than treating it as cessation, and most quoted figures do not say which they used.

Two censoring rules are then required. The 2020 and 2021 seasons were interrupted by public-health restrictions rather than by market conditions, and cohorts spanning them must either exclude those cycles from the survival clock or report both treatments. Events younger than the observation window are right-censored and contribute exposure time without contributing an outcome.

Where the roster could come from

Four sources can be combined. Trade compilations record larger events with reported attendance, unevenly across the period.¹ Association records document closures and cancellations for their membership.²³ Contemporaneous lineup announcements and archived event sites establish that an edition occurred in a given year. Local permitting and municipal records establish site use for events on public land, and are retrievable for many jurisdictions.

The combination is labour-intensive and imperfect: it will systematically under-record small and short-lived events, which biases any survival estimate upward. That bias should be stated with the estimate rather than corrected by assumption.

What Live Index will publish, and when

A bounded cohort — a defined vintage range, a stated capacity floor and a named set of metropolitan areas — is achievable and is on the work programme. When it is published, it will carry the roster itself, the survival state of each event with its source, and the rate under each treatment of transformation and pandemic censoring. Until then, Live Index states no survival rate and treats those in circulation as unsourced.